fix(admin): keep header-style tiles from overflowing their cards (stable) (#1423)

Apply the header-style overflow fix to stable: size the grids from their container width and wrap long translated labels within each card.

.claudedocs/scans/security-20250713.md

@@ -1,286 +0,0 @@
-# Security Scan Report - Wedding Photo Sharing Application
-**Date**: July 13, 2025  
-**Scanner**: Claude Security Audit with --security --validate flags  
-**Overall Risk Level**: MEDIUM-HIGH  
-
-## Executive Summary
-
-The wedding photo sharing application demonstrates strong security fundamentals with comprehensive input validation, proper authentication mechanisms, and good file security practices. However, several critical issues require immediate attention, particularly around hardcoded secrets, token storage, and Content Security Policy configuration.
-
-### Security Score: 6.5/10
-
-**Strengths**: Excellent input validation, parameterized queries, file security, rate limiting  
-**Critical Issues**: Hardcoded JWT secrets, localStorage token storage, weak CSP, console logging in production
-
----
-
-## 🔴 CRITICAL FINDINGS (Immediate Action Required)
-
-### 1. Hardcoded JWT Secret in Development
-- **Location**: Backend `.env` file
-- **Risk**: Token forgery, authentication bypass
-- **Impact**: Complete authentication compromise
-- **Remediation**: 
-  ```bash
-  # Generate secure secret
-  openssl rand -base64 32
-  # Never commit to repository
-  echo ".env" >> .gitignore
-  ```
-
-### 2. Gallery Tokens in localStorage
-- **Location**: Frontend `api.ts` and auth contexts
-- **Risk**: XSS token theft
-- **Impact**: Gallery access compromise
-- **Remediation**: Move to httpOnly cookies:
-  ```typescript
-  Cookies.set(`gallery_token_${slug}`, token, {
-    httpOnly: true,
-    secure: true,
-    sameSite: 'strict'
-  });
-  ```
-
-### 3. Weak Content Security Policy
-- **Location**: Frontend `nginx.conf`
-- **Risk**: XSS, code injection
-- **Current**: `unsafe-inline` and `unsafe-eval` allowed
-- **Remediation**: Implement strict CSP (see detailed recommendations below)
-
----
-
-## 🟠 HIGH SEVERITY FINDINGS
-
-### 1. Console Logging in Production
-- **Locations**: 61 instances across frontend
-- **Risk**: Information disclosure
-- **Impact**: Leaking sensitive data, debugging info
-- **Remediation**: Implement environment-aware logging
-
-### 2. Token Revocation Vulnerability
-- **Location**: Backend `tokenRevocation.js`
-- **Risk**: Token manipulation
-- **Impact**: Bypass revocation checks
-- **Remediation**: Verify token signature before decoding
-
-### 3. Source Maps in Production
-- **Location**: Frontend build configuration
-- **Risk**: Source code exposure
-- **Impact**: Reveals application structure
-- **Remediation**: Disable in production builds
-
-### 4. Missing Security Headers
-- **Location**: nginx configuration
-- **Missing**: HSTS, Permissions-Policy
-- **Impact**: Various client-side attacks
-- **Remediation**: Add comprehensive security headers
-
----
-
-## 🟡 MEDIUM SEVERITY FINDINGS
-
-### 1. Rate Limiting Bypass Potential
-- **Location**: Backend rate limiter
-- **Risk**: DoS attacks
-- **Current**: JWT validation in rate limiter
-- **Remediation**: Use IP-based limiting only
-
-### 2. Incomplete SQL Injection Protection
-- **Location**: Complex dashboard queries
-- **Risk**: Potential injection in edge cases
-- **Current**: Mostly parameterized
-- **Remediation**: Use query builder exclusively
-
-### 3. Session Management
-- **Issue**: No gallery token invalidation on password change
-- **Risk**: Persistent access after compromise
-- **Remediation**: Implement token revocation
-
-### 4. Path Traversal in Gallery Slugs
-- **Location**: Frontend gallery routes
-- **Risk**: Directory traversal attempts
-- **Remediation**: Validate and sanitize slugs
-
----
-
-## 🟢 LOW SEVERITY FINDINGS
-
-### 1. Verbose Error Messages
-- **Location**: Multiple API endpoints
-- **Risk**: Information disclosure
-- **Remediation**: Generic client errors, detailed server logs
-
-### 2. Weak Gallery Passwords
-- **Current**: zxcvbn score 2/4 allowed
-- **Risk**: Brute force attacks
-- **Remediation**: Increase to score 3/4
-
-### 3. Missing File Size Validation
-- **Location**: Frontend upload components
-- **Risk**: DoS via large uploads
-- **Remediation**: Add client-side size checks
-
----
-
-## ✅ SECURITY STRENGTHS
-
-### Authentication & Authorization
-- JWT with proper expiration (24h/7d)
-- Token type validation
-- IP tracking and validation
-- Password change detection
-- Token revocation system
-- Bcrypt with 12 rounds
-- zxcvbn password strength checking
-
-### Input Validation & SQL Security
-- express-validator on all endpoints
-- Parameterized queries via Knex
-- SQL injection protection utilities
-- Path traversal prevention
-- Comprehensive input sanitization
-
-### File Security
-- Magic number verification
-- MIME type validation
-- Safe filename generation
-- Directory traversal protection
-- File extension whitelist
-
-### Rate Limiting & DoS Protection
-- General: 100 req/15min
-- Auth endpoints: 5 req/15min
-- Account lockout after failed attempts
-- Suspicious activity detection
-
-### Frontend Security
-- React's built-in XSS protection
-- DOMPurify for HTML content
-- No eval() or innerHTML usage
-- Proper error boundaries
-- ReCAPTCHA integration
-
----
-
-## 📊 DEPENDENCY ANALYSIS
-
-### Current Status
-- **Backend**: 0 vulnerabilities (691 packages)
-- **Frontend**: 0 vulnerabilities (434 packages)
-
-### Recommended Updates
-1. **bcrypt** 5.1.1 → 6.0.0 (performance, compatibility)
-2. **helmet** 7.2.0 → 8.1.0 (new security features)
-3. **@tiptap** 2.x → 3.x (security improvements)
-
-### Supply Chain Assessment
-- All major dependencies from trusted sources
-- No typosquatting detected
-- Regular maintenance observed
-- MIT/ISC/Apache licenses only
-
----
-
-## 🛠️ REMEDIATION PLAN
-
-### Phase 1: Critical (Within 24 hours)
-1. Replace hardcoded JWT secret with secure random value
-2. Move gallery tokens from localStorage to httpOnly cookies
-3. Implement strict CSP without unsafe-eval
-4. Remove or wrap console.log statements
-
-### Phase 2: High Priority (Within 1 week)
-1. Disable source maps in production
-2. Add missing security headers (HSTS, Permissions-Policy)
-3. Fix token revocation vulnerability
-4. Update critical dependencies (bcrypt, helmet)
-
-### Phase 3: Medium Priority (Within 1 month)
-1. Implement comprehensive logging strategy
-2. Add gallery slug validation
-3. Enhance rate limiting logic
-4. Implement session invalidation on password change
-
-### Phase 4: Ongoing
-1. Weekly dependency scanning
-2. Implement security testing in CI/CD
-3. Regular penetration testing
-4. Security awareness training
-
----
-
-## 🔒 RECOMMENDED CSP CONFIGURATION
-
-```nginx
-add_header Content-Security-Policy "
-  default-src 'self';
-  script-src 'self' 'nonce-{RANDOM}' https://www.google.com/recaptcha/ https://www.gstatic.com/recaptcha/;
-  style-src 'self' 'unsafe-inline';
-  img-src 'self' data: blob: https:;
-  font-src 'self';
-  connect-src 'self' https://analytics.domain.com;
-  frame-src https://www.google.com/recaptcha/;
-  object-src 'none';
-  base-uri 'self';
-  form-action 'self';
-  frame-ancestors 'none';
-  upgrade-insecure-requests;
-" always;
-```
-
----
-
-## 🚀 SECURITY IMPROVEMENTS ROADMAP
-
-### Immediate Implementation
-```bash
-# 1. Generate secure secrets
-openssl rand -base64 32 > jwt-secret.txt
-
-# 2. Update dependencies
-cd backend && npm install bcrypt@^6.0.0 helmet@^8.1.0
-cd ../frontend && npm update
-
-# 3. Add security scanning
-npm install -D npm-audit-resolver
-```
-
-### CI/CD Integration
-```yaml
-# Add to CI pipeline
-- name: Security Scan
-  run: |
-    npm audit --audit-level=moderate
-    npm run test:security
-```
-
-### Monitoring & Alerting
-1. Implement fail2ban for repeated auth failures
-2. Set up log analysis for suspicious patterns
-3. Configure alerts for security events
-4. Regular vulnerability scanning
-
----
-
-## 📋 COMPLIANCE CHECKLIST
-
-- [ ] OWASP Top 10 addressed
-- [ ] GDPR compliance (data minimization, right to erasure)
-- [ ] Security headers implemented
-- [ ] Dependency scanning automated
-- [ ] Incident response plan documented
-- [ ] Security documentation maintained
-- [ ] Regular security reviews scheduled
-
----
-
-## 🎯 CONCLUSION
-
-The wedding photo sharing application has a solid security foundation with excellent input validation and authentication mechanisms. However, operational security practices need immediate attention. The critical issues around secret management and token storage must be addressed before production deployment.
-
-Implementing the recommended fixes will raise the security score from 6.5/10 to approximately 8.5/10, providing a robust and secure platform for wedding photo sharing.
-
----
-
-*Generated by Claude Security Scanner v1.0*  
-*Next scan recommended: After Phase 1 remediation completion*

.drone.yml

@@ -1,77 +0,0 @@
-kind: pipeline
-type: docker
-name: default
-
-steps:
-  # Build Backend Docker Image
-  - name: build-backend
-    image: plugins/docker
-    settings:
-      repo: registry.local.nothaft.cloud/picpeak-backend
-      tags: 
-        - latest
-        - ${DRONE_COMMIT_SHA:0:8}
-        - ${DRONE_BRANCH}-latest
-      dockerfile: backend/Dockerfile
-      context: backend/
-      registry: registry.local.nothaft.cloud
-      build_args:
-        - VERSION=${DRONE_TAG:-dev}
-      
-  # Build Frontend Docker Image
-  - name: build-frontend
-    image: plugins/docker
-    settings:
-      repo: registry.local.nothaft.cloud/picpeak-frontend
-      tags: 
-        - latest
-        - ${DRONE_COMMIT_SHA:0:8}
-        - ${DRONE_BRANCH}-latest
-      dockerfile: frontend/Dockerfile
-      context: frontend/
-      registry: registry.local.nothaft.cloud
-      build_args:
-        - VERSION=${DRONE_TAG:-dev}
-        - VITE_API_URL=${VITE_API_URL:-/api}
-
-trigger:
-  branch:
-    - main
-    - develop
-  event:
-    - push
-    - pull_request
-
----
-kind: pipeline
-type: docker
-name: release
-
-steps:
-  # Build Backend Release
-  - name: build-backend-release
-    image: plugins/docker
-    settings:
-      repo: registry.local.nothaft.cloud/picpeak-backend
-      tags:
-        - ${DRONE_TAG}
-        - latest
-      dockerfile: backend/Dockerfile
-      context: backend/
-      registry: registry.local.nothaft.cloud
-      
-  # Build Frontend Release
-  - name: build-frontend-release
-    image: plugins/docker
-    settings:
-      repo: registry.local.nothaft.cloud/picpeak-frontend
-      tags:
-        - ${DRONE_TAG}
-        - latest
-      dockerfile: frontend/Dockerfile
-      context: frontend/
-      registry: registry.local.nothaft.cloud
-
-trigger:
-  event:
-    - tag

.drone.yml.full

@@ -1,280 +0,0 @@
-kind: pipeline
-type: docker
-name: default
-
-trigger:
-  branch:
-    - main
-    - develop
-    - feature/*
-  event:
-    - push
-    - pull_request
-    - tag
-
-volumes:
-  - name: docker
-    host:
-      path: /var/run/docker.sock
-
-steps:
-  # Frontend Tests
-  - name: frontend-test
-    image: node:18-alpine
-    commands:
-      - cd frontend
-      - npm ci --legacy-peer-deps
-      - npm run lint
-      - npm run build
-    when:
-      event:
-        - push
-        - pull_request
-
-  # Backend Tests
-  - name: backend-test
-    image: node:18-alpine
-    commands:
-      - cd backend
-      - npm ci
-      - npm run lint
-      - npm test
-    environment:
-      NODE_ENV: test
-      JWT_SECRET: test-secret
-    when:
-      event:
-        - push
-        - pull_request
-
-  # Build Frontend Docker Image
-  - name: build-frontend
-    image: plugins/docker
-    settings:
-      repo: registry.local.nothaft.cloud/wedding-photo-sharing-frontend
-      tags:
-        - latest
-        - ${DRONE_COMMIT_SHA:0:8}
-        - ${DRONE_TAG}
-      dockerfile: frontend/Dockerfile
-      context: frontend
-      registry: registry.local.nothaft.cloud
-    when:
-      branch:
-        - main
-      event:
-        - push
-        - tag
-
-  # Build Backend Docker Image
-  - name: build-backend
-    image: plugins/docker
-    settings:
-      repo: registry.local.nothaft.cloud/wedding-photo-sharing-backend
-      tags:
-        - latest
-        - ${DRONE_COMMIT_SHA:0:8}
-        - ${DRONE_TAG}
-      dockerfile: backend/Dockerfile
-      context: backend
-      registry: registry.local.nothaft.cloud
-    when:
-      branch:
-        - main
-      event:
-        - push
-        - tag
-
-  # Security Scan
-  - name: security-scan
-    image: aquasec/trivy:latest
-    commands:
-      - trivy image --exit-code 0 --no-progress registry.local.nothaft.cloud/wedding-photo-sharing-frontend:${DRONE_COMMIT_SHA:0:8}
-      - trivy image --exit-code 0 --no-progress registry.local.nothaft.cloud/wedding-photo-sharing-backend:${DRONE_COMMIT_SHA:0:8}
-    environment:
-      DOCKER_HOST: tcp://docker:2375
-    volumes:
-      - name: docker
-        path: /var/run/docker.sock
-    when:
-      branch:
-        - main
-      event:
-        - push
-
-  # Deploy to Staging
-  - name: deploy-staging
-    image: alpine:latest
-    environment:
-      SWARM_HOST:
-        from_secret: staging_swarm_host
-      SWARM_USER:
-        from_secret: staging_swarm_user
-      SWARM_KEY:
-        from_secret: staging_swarm_key
-      REGISTRY_URL:
-        from_secret: docker_registry
-      VERSION: ${DRONE_COMMIT_SHA:0:8}
-    commands:
-      - apk add --no-cache openssh-client
-      - mkdir -p ~/.ssh
-      - echo "$SWARM_KEY" > ~/.ssh/id_rsa
-      - chmod 600 ~/.ssh/id_rsa
-      - ssh-keyscan -H $SWARM_HOST >> ~/.ssh/known_hosts
-      - |
-        ssh $SWARM_USER@$SWARM_HOST << EOF
-          cd /opt/wedding-photo-sharing
-          export REGISTRY_URL=registry.local.nothaft.cloud
-          export VERSION=$VERSION
-          docker stack deploy -c deploy/docker-stack.yml wedding-photo-sharing
-        EOF
-    when:
-      branch:
-        - develop
-      event:
-        - push
-
-  # Deploy to Production
-  - name: deploy-production
-    image: alpine:latest
-    environment:
-      SWARM_HOST:
-        from_secret: prod_swarm_host
-      SWARM_USER:
-        from_secret: prod_swarm_user
-      SWARM_KEY:
-        from_secret: prod_swarm_key
-      REGISTRY_URL:
-        from_secret: docker_registry
-      VERSION: ${DRONE_TAG:-latest}
-    commands:
-      - apk add --no-cache openssh-client
-      - mkdir -p ~/.ssh
-      - echo "$SWARM_KEY" > ~/.ssh/id_rsa
-      - chmod 600 ~/.ssh/id_rsa
-      - ssh-keyscan -H $SWARM_HOST >> ~/.ssh/known_hosts
-      - |
-        ssh $SWARM_USER@$SWARM_HOST << EOF
-          cd /opt/wedding-photo-sharing
-          export REGISTRY_URL=registry.local.nothaft.cloud
-          export VERSION=$VERSION
-          
-          # Backup database before deployment
-          docker exec \$(docker ps -q -f name=wedding-photo-sharing_db) pg_dump -U postgres wedding_photo_sharing > /backup/db-backup-\$(date +%Y%m%d-%H%M%S).sql
-          
-          # Deploy stack
-          docker stack deploy -c deploy/docker-stack.yml wedding-photo-sharing --with-registry-auth
-          
-          # Wait for services to be ready
-          sleep 30
-          
-          # Run migrations if needed
-          docker exec \$(docker ps -q -f name=wedding-photo-sharing_backend) npm run migrate
-        EOF
-    when:
-      event:
-        - tag
-
-  # Health Check
-  - name: health-check
-    image: alpine:latest
-    commands:
-      - apk add --no-cache curl
-      - sleep 30
-      - curl -f https://${FRONTEND_HOST}/health || exit 1
-      - curl -f https://${BACKEND_HOST}/api/health || exit 1
-    when:
-      branch:
-        - main
-      event:
-        - push
-        - tag
-
-  # Notification - Success
-  - name: notify-success
-    image: plugins/slack
-    settings:
-      webhook:
-        from_secret: slack_webhook
-      channel: deployments
-      template: |
-        ✅ *Build {{build.number}} succeeded* for {{repo.name}}
-        
-        Branch: {{build.branch}}
-        Commit: {{build.commit}}
-        Author: {{build.author}}
-        
-        {{#if build.tag}}
-        🏷️ Tag: {{build.tag}}
-        🚀 Deployed to *PRODUCTION*
-        {{else}}
-        📦 Deployed to *{{build.branch}}*
-        {{/if}}
-        
-        🔗 {{build.link}}
-    when:
-      status:
-        - success
-
-  # Notification - Failure
-  - name: notify-failure
-    image: plugins/slack
-    settings:
-      webhook:
-        from_secret: slack_webhook
-      channel: deployments
-      template: |
-        ❌ *Build {{build.number}} failed* for {{repo.name}}
-        
-        Branch: {{build.branch}}
-        Commit: {{build.commit}}
-        Author: {{build.author}}
-        
-        🔗 {{build.link}}
-    when:
-      status:
-        - failure
-
----
-kind: pipeline
-type: docker
-name: rollback
-
-trigger:
-  event:
-    - rollback
-
-steps:
-  - name: rollback-production
-    image: alpine:latest
-    environment:
-      SWARM_HOST:
-        from_secret: prod_swarm_host
-      SWARM_USER:
-        from_secret: prod_swarm_user
-      SWARM_KEY:
-        from_secret: prod_swarm_key
-      REGISTRY_URL:
-        from_secret: docker_registry
-    commands:
-      - apk add --no-cache openssh-client
-      - mkdir -p ~/.ssh
-      - echo "$SWARM_KEY" > ~/.ssh/id_rsa
-      - chmod 600 ~/.ssh/id_rsa
-      - ssh-keyscan -H $SWARM_HOST >> ~/.ssh/known_hosts
-      - |
-        ssh $SWARM_USER@$SWARM_HOST << EOF
-          cd /opt/wedding-photo-sharing
-          export REGISTRY_URL=registry.local.nothaft.cloud
-          export VERSION=${DRONE_ROLLBACK_TO}
-          
-          # Deploy previous version
-          docker stack deploy -c deploy/docker-stack.yml wedding-photo-sharing --with-registry-auth
-        EOF
-
----
-kind: secret
-name: slack_webhook
-get:
-  path: drone/slack
-  name: webhook

.env.example

@@ -1,34 +1,217 @@
-# Environment Configuration Template
-# Copy this file to .env and adjust values for your environment
+# PicPeak Environment Configuration
+# Copy this file to .env and update with your values
 
-# Development: Use docker-compose.dev.yml
-# Production: Use docker-compose.prod.yml with .env.production.example
+# Environment
+NODE_ENV=production
 
-# JWT Secret (CRITICAL for production)
-# Generate with: openssl rand -base64 32
-JWT_SECRET=dev-secret-change-in-production
+# JWT Secret — OPTIONAL. Leave unset and it is auto-generated on first run
+# (Docker: the secrets-init service writes it to a private volume and reuses it
+# across restarts). Set it explicitly only to pin your own value.
+# Generate one with: openssl rand -base64 64
+#JWT_SECRET=your_very_long_random_jwt_secret_here
 
-# Application URLs
-ADMIN_URL=http://localhost:3005
-FRONTEND_URL=http://localhost:3005
+# Auth cookie Secure flag
+#   unset  - default: follows NODE_ENV (production=true, dev=false)
+#   true   - always set Secure (HTTPS-only cookies; breaks plain-HTTP access)
+#   false  - never set Secure (allows HTTP; cookies not protected on HTTPS)
+#   auto   - decide per request: Secure on HTTPS, not on HTTP
+#
+# Use COOKIE_SECURE=auto if your deployment is reachable over both HTTPS
+# (via reverse proxy like Nginx Proxy Manager, Traefik, Caddy) AND plain
+# HTTP (e.g. LAN access at http://192.168.x.x:3010). The backend reads
+# req.secure from Express, which respects the X-Forwarded-Proto header
+# when the proxy is in the trust list.
+#
+# Requirements for auto mode:
+#   1. Your reverse proxy MUST send X-Forwarded-Proto: https on HTTPS
+#      requests. Standard configs for NPM/Traefik/Caddy do this by default.
+#   2. The proxy must be on a trusted IP range. By default PicPeak trusts
+#      loopback and private networks (127.0.0.1, 10.x, 172.16-31.x,
+#      192.168.x, link-local). Proxies outside those ranges need custom
+#      trust proxy configuration.
+# COOKIE_SECURE=auto
 
-# Database Configuration
-# SQLite is used for development by default
-# For production PostgreSQL config, see .env.production.example
-DATABASE_CLIENT=sqlite3
-DATABASE_PATH=./data/photo_sharing.db
+# Cookie SameSite attribute (Lax | Strict | None). Default: Lax
+# COOKIE_SAMESITE=Lax
+
+# Cookie Domain — set this if serving auth cookies across subdomains.
+# Leave unset for same-origin setups.
+# COOKIE_DOMAIN=.example.com
+
+# Database Configuration (PostgreSQL)
+DATABASE_CLIENT=pg
+DB_USER=picpeak
+# DB_PASSWORD — OPTIONAL. Leave unset and it is auto-generated on first run
+# (Docker). Set it explicitly to pin your own, e.g. for an external database.
+# IMPORTANT: Avoid $ character in passwords - Docker Compose interprets it as variable substitution
+# If you must use $, escape it as $$ (e.g., Pass$$word instead of Pass$word)
+#DB_PASSWORD=your_secure_postgres_password_here
+DB_NAME=picpeak_prod
+
+# Redis Configuration
+# REDIS_PASSWORD — OPTIONAL. Leave unset and it is auto-generated on first run (Docker).
+# IMPORTANT: Same warning applies - avoid $ or escape as $$
+#REDIS_PASSWORD=your_secure_redis_password_here
+
+# Admin Account (initial setup) — OPTIONAL
+# Leave these unset (default) to create your admin IN THE BROWSER on first run:
+# open /admin and PicPeak shows a setup screen. The one-time setup token is
+# written to data/SETUP_TOKEN with mode 0600 — read it with
+# `docker compose exec backend cat /app/data/SETUP_TOKEN`. It is NOT logged
+# unless that write fails, so it never sits in `docker logs`.
+# Set ADMIN_PASSWORD to auto-create the admin on first boot instead (legacy;
+# credentials written to data/ADMIN_CREDENTIALS.txt).
+#ADMIN_USERNAME=admin
+#ADMIN_EMAIL=admin@yourdomain.com
+#ADMIN_PASSWORD=your_secure_admin_password_here
 
 # Email Configuration
-# Development: Uses Mailhog (included in docker-compose.dev.yml)
-# Production: Configure real SMTP server
-SMTP_HOST=mailhog
-SMTP_PORT=1025
+# For Gmail: use app-specific password
+# For SendGrid: SMTP_USER=apikey, SMTP_PASS=your-api-key
+SMTP_HOST=smtp.gmail.com
+SMTP_PORT=587
 SMTP_SECURE=false
-SMTP_USER=
-SMTP_PASS=
-EMAIL_FROM=noreply@localhost
+SMTP_USER=your-email@gmail.com
+SMTP_PASS=your-app-specific-password
+EMAIL_FROM=noreply@yourdomain.com
 
-# Optional: Umami Analytics
-UMAMI_URL=
-UMAMI_WEBSITE_ID=
-UMAMI_HASH_SALT=
+# Application URLs
+# Use full origin with scheme, no trailing slash.
+# Admin UI is served by the frontend at /admin.
+FRONTEND_URL=https://yourdomain.com
+ADMIN_URL=https://yourdomain.com
+
+# Static HTML title + description used for social link previews when the
+# fetcher doesn't trigger the per-event OG endpoint — most notably the
+# WhatsApp Business API and various 3rd-party preview-service caches
+# (#521). Set these to your brand so link previews aren't generic.
+# Substituted into index.html at frontend-container start, so changes
+# take effect on the next `docker compose up -d frontend` — no rebuild
+# required.
+BRAND_TITLE=PicPeak
+BRAND_DESCRIPTION=Photo gallery shared with PicPeak.
+
+# API URL for email assets (logos, images in notification emails)
+# This must be the publicly accessible URL where email recipients can load images.
+# If not set, defaults to http://localhost:3001 which will show broken images in emails.
+API_URL=https://yourdomain.com/api
+
+# Frontend API base
+# For pre-built images and production behind a reverse proxy, keep '/api'.
+# If you rebuild the frontend yourself, you may set a full URL at build time.
+VITE_API_URL=/api
+
+# Port Configuration (optional)
+# BACKEND_PORT=3001
+# FRONTEND_PORT=3000
+# DB_PORT=5432
+# REDIS_PORT=6379
+
+# Release Channel
+# Options: 'stable' (default), 'beta', or specific version like 'v2.3.0'
+# 'stable' uses the :stable tag (same as :latest on main)
+# 'beta' uses the :beta tag for pre-release versions
+PICPEAK_CHANNEL=stable
+
+# Update Check Configuration
+# Set to 'false' to disable update notifications in admin UI
+UPDATE_CHECK_ENABLED=true
+
+# Timezone
+TZ=UTC
+
+# Analytics (Optional - Umami)
+VITE_UMAMI_URL=
+VITE_UMAMI_WEBSITE_ID=
+VITE_UMAMI_SHARE_URL=
+
+# Storage variables (host paths)
+# These control where data is stored on the host. Defaults are local folders.
+APP_STORAGE=./storage
+APP_DATA=./data
+LOGS=./logs
+
+# ─── Storage Backend ────────────────────────────────────────────────────────
+# PicPeak can store photos, thumbnails and archive zips on the local filesystem
+# (default) or on any S3-compatible object store (AWS S3, MinIO, Cloudflare R2,
+# Backblaze B2, Wasabi, DigitalOcean Spaces, …).
+#
+# STORAGE_BACKEND=local (default)
+#   Uses STORAGE_PATH on the local filesystem. Backwards compatible — every
+#   existing deployment keeps working unchanged.
+#
+# STORAGE_BACKEND=s3
+#   Reads STORAGE_S3_* below. Auto-import via the filesystem watcher is
+#   disabled in this mode (S3 has no inotify) — every photo must enter via the
+#   admin upload UI/API. Run `node backend/scripts/migrate-storage.js` to copy
+#   existing local content to S3 before flipping the env.
+#
+# STORAGE_BACKEND=local
+#
+# STORAGE_S3_BUCKET=picpeak
+# STORAGE_S3_REGION=us-east-1
+# STORAGE_S3_ACCESS_KEY=AKIAxxxxxxxxxxxxxxxx
+# STORAGE_S3_SECRET_KEY=xxxxxxxxxxxxxxxxxxxxxxxx
+# Custom endpoint — set this for MinIO / R2 / B2 / Spaces. Leave unset for AWS.
+# STORAGE_S3_ENDPOINT=https://s3.us-west-002.backblazeb2.com
+# Optional namespace prefix inside the bucket — useful for multi-deployment buckets.
+# STORAGE_S3_PREFIX=picpeak
+# STORAGE_S3_FORCE_PATH_STYLE=false   # MinIO needs true; auto-on when endpoint is set
+# STORAGE_S3_SSL=true
+#
+# Minimum IAM policy (AWS S3) for the bucket above:
+#   {
+#     "Version": "2012-10-17",
+#     "Statement": [{
+#       "Effect": "Allow",
+#       "Action": [
+#         "s3:GetObject", "s3:PutObject", "s3:DeleteObject",
+#         "s3:ListBucket", "s3:GetBucketLocation"
+#       ],
+#       "Resource": [
+#         "arn:aws:s3:::picpeak",
+#         "arn:aws:s3:::picpeak/*"
+#       ]
+#     }]
+#   }
+#
+# EXTERNAL_MEDIA_ROOT (above) always lives on the local filesystem regardless
+# of STORAGE_BACKEND — reference-mode galleries are not migrated to S3 in v1.
+
+# ─── Outbound Webhooks (#327) ────────────────────────────────────────────────
+# PicPeak POSTs event/photo lifecycle notifications to URLs you configure
+# under Settings → Webhooks. Each delivery is signed HMAC-SHA256 with a
+# per-webhook secret in the X-PicPeak-Signature header.
+#
+# WEBHOOK_ALLOW_PRIVATE_URLS (default: false)
+#   Block URLs resolving to private IPs / loopback / .local etc. as an
+#   SSRF mitigation. Set to "true" ONLY in dev when your receiver is on
+#   the same docker network or localhost. Production deployments must
+#   leave this OFF.
+# WEBHOOK_ALLOW_PRIVATE_URLS=false
+#
+# WEBHOOK_DELIVERY_INTERVAL_MS (default: 5000)
+#   How often the worker polls webhook_deliveries for pending rows.
+# WEBHOOK_DELIVERY_INTERVAL_MS=5000
+#
+# WEBHOOK_DELIVERY_CONCURRENCY (default: 5)
+#   Maximum in-flight deliveries per worker tick. One slow consumer can
+#   monopolize all 5 slots — bump this if your receivers are slow OR ship
+#   a separate webhook-only deployment.
+# WEBHOOK_DELIVERY_CONCURRENCY=5
+#
+# WEBHOOK_HTTP_TIMEOUT_MS (default: 10000)
+#   Per-request timeout. Beyond this, the delivery is recorded as a
+#   network error and retried.
+# WEBHOOK_HTTP_TIMEOUT_MS=10000
+#
+# WEBHOOK_MAX_ATTEMPTS (default: 5)
+#   Total attempts before a delivery is marked failed. Backoff between
+#   attempts is exponential: 1m, 5m, 30m, 2h, 12h.
+# WEBHOOK_MAX_ATTEMPTS=5
+
+# Note on FRONTEND_API_URL (documentation only):
+# When using pre-built frontend images, runtime env vars cannot override the built JS.
+# Do NOT rely on FRONTEND_API_URL in Compose. Instead, keep VITE_API_URL=/api and
+# let the frontend Nginx proxy /api to the backend. Only if you rebuild the frontend
+# should you change VITE_API_URL at build time.

.env.production.example

@@ -1,32 +0,0 @@
-# Production Environment Configuration Template
-# Copy this file to .env and fill in your values
-
-# Application URLs
-ADMIN_URL=https://yourdomain.com
-FRONTEND_URL=https://yourdomain.com
-
-# Security - CRITICAL: Generate a secure random JWT secret
-# You can generate one with: openssl rand -base64 32
-JWT_SECRET=your-secure-random-jwt-secret-here
-
-# Database Configuration (PostgreSQL)
-DB_USER=picpeak
-DB_PASSWORD=your-secure-database-password
-DB_NAME=picpeak
-
-# Email Configuration
-SMTP_HOST=smtp.gmail.com
-SMTP_PORT=587
-SMTP_SECURE=true
-SMTP_USER=your-email@gmail.com
-SMTP_PASS=your-app-password
-EMAIL_FROM=noreply@yourdomain.com
-
-# Umami Analytics (Optional)
-UMAMI_URL=https://analytics.yourdomain.com
-UMAMI_WEBSITE_ID=your-website-id
-UMAMI_HASH_SALT=your-random-hash-salt
-
-# First Admin User (for initial setup)
-# Run: docker-compose exec backend node scripts/create-admin.js --email admin@yourdomain.com
-ADMIN_EMAIL=admin@yourdomain.com

.gitea/workflows/test.yml

@@ -1,52 +0,0 @@
-name: Test and Lint
-
-on:
-  push:
-    branches: [ main, develop ]
-  pull_request:
-    branches: [ main ]
-
-jobs:
-  backend-test:
-    runs-on: ubuntu-latest
-    steps:
-      - uses: actions/checkout@v3
-      
-      - name: Setup Node.js
-        uses: actions/setup-node@v3
-        with:
-          node-version: '18'
-      
-      - name: Install backend dependencies
-        working-directory: ./backend
-        run: npm ci
-      
-      - name: Run backend linting
-        working-directory: ./backend
-        run: npm run lint || true  # Continue on lint errors for now
-      
-      - name: Run backend tests
-        working-directory: ./backend
-        run: npm test || true  # Continue on test failures for now
-  
-  frontend-test:
-    runs-on: ubuntu-latest
-    steps:
-      - uses: actions/checkout@v3
-      
-      - name: Setup Node.js
-        uses: actions/setup-node@v3
-        with:
-          node-version: '18'
-      
-      - name: Install frontend dependencies
-        working-directory: ./frontend
-        run: npm ci --legacy-peer-deps
-      
-      - name: Run frontend linting
-        working-directory: ./frontend
-        run: npm run lint || true  # Continue on lint errors for now
-      
-      - name: Build frontend
-        working-directory: ./frontend
-        run: npm run build

.gitea/workflows/version-and-release.yml

@@ -1,88 +0,0 @@
-name: Version and Release
-
-on:
-  push:
-    branches: [ main ]
-    paths-ignore:
-      - '**.md'
-      - '.gitea/**'
-      - '.drone.yml'
-
-jobs:
-  version-bump:
-    runs-on: ubuntu-latest
-    outputs:
-      new_version: ${{ steps.version.outputs.new_version }}
-      version_changed: ${{ steps.version.outputs.version_changed }}
-    steps:
-      - uses: actions/checkout@v3
-        with:
-          fetch-depth: 0
-          token: ${{ secrets.GITEA_TOKEN || github.token }}
-      
-      - name: Setup Node.js
-        uses: actions/setup-node@v3
-        with:
-          node-version: '18'
-      
-      - name: Configure Git
-        run: |
-          git config --global user.name 'Gitea Actions Bot'
-          git config --global user.email 'actions@gitea.local'
-      
-      - name: Bump version
-        id: version
-        run: |
-          # Get current version from backend package.json
-          CURRENT_VERSION=$(node -p "require('./backend/package.json').version")
-          echo "Current version: $CURRENT_VERSION"
-          
-          # Split version into parts
-          IFS='.' read -r -a version_parts <<< "$CURRENT_VERSION"
-          MAJOR="${version_parts[0]}"
-          MINOR="${version_parts[1]}"
-          PATCH="${version_parts[2]}"
-          
-          # Increment patch version
-          NEW_PATCH=$((PATCH + 1))
-          NEW_VERSION="$MAJOR.$MINOR.$NEW_PATCH"
-          
-          echo "New version: $NEW_VERSION"
-          echo "new_version=$NEW_VERSION" >> $GITHUB_OUTPUT
-          
-          # Update version in package.json files
-          cd backend && npm version $NEW_VERSION --no-git-tag-version
-          cd ../frontend && npm version $NEW_VERSION --no-git-tag-version
-          cd ..
-          
-          # Check if there are changes
-          if [[ -n $(git status -s) ]]; then
-            echo "version_changed=true" >> $GITHUB_OUTPUT
-          else
-            echo "version_changed=false" >> $GITHUB_OUTPUT
-          fi
-      
-      - name: Commit version bump
-        if: steps.version.outputs.version_changed == 'true'
-        run: |
-          git add backend/package.json backend/package-lock.json
-          git add frontend/package.json frontend/package-lock.json
-          git commit -m "chore: bump version to ${{ steps.version.outputs.new_version }}"
-          git push
-      
-      - name: Create Git tag
-        if: steps.version.outputs.version_changed == 'true'
-        run: |
-          git tag -a "v${{ steps.version.outputs.new_version }}" -m "Release v${{ steps.version.outputs.new_version }}"
-          git push origin "v${{ steps.version.outputs.new_version }}"
-  
-  trigger-drone:
-    needs: version-bump
-    if: needs.version-bump.outputs.version_changed == 'true'
-    runs-on: ubuntu-latest
-    steps:
-      - name: Trigger Drone Build
-        run: |
-          echo "Version bumped to ${{ needs.version-bump.outputs.new_version }}"
-          echo "Drone will automatically trigger on the new tag"
-          # Drone CI will automatically trigger on the tag push event

.github/FUNDING.yml

@@ -0,0 +1,4 @@
+# These are supported funding model platforms
+# https://docs.github.com/en/repositories/managing-your-repositorys-settings-and-features/customizing-your-repository/displaying-a-sponsor-button-in-your-repository
+
+buy_me_a_coffee: theluap

.github/ISSUE_TEMPLATE/bug_report.md

@@ -0,0 +1,47 @@
+---
+name: Bug report
+about: Create a report to help us improve PicPeak
+title: '[BUG] '
+labels: 'bug'
+assignees: ''
+
+---
+
+**Describe the bug**
+A clear and concise description of what the bug is.
+
+**To Reproduce**
+Steps to reproduce the behavior:
+1. Go to '...'
+2. Click on '....'
+3. Scroll down to '....'
+4. See error
+
+**Expected behavior**
+A clear and concise description of what you expected to happen.
+
+**Screenshots**
+If applicable, add screenshots to help explain your problem.
+
+**Environment (please complete the following information):**
+ - OS: [e.g. Ubuntu 22.04]
+ - Browser: [e.g. Chrome 120, Safari 17]
+ - PicPeak Version: [e.g. 1.0.22]
+ - Deployment Method: [e.g. Docker Compose, Manual]
+ - Database: [e.g. PostgreSQL 15, SQLite]
+
+**Logs**
+Please include relevant logs:
+```
+# Backend logs
+docker-compose logs backend | tail -50
+
+# Frontend console errors
+[paste any browser console errors]
+```
+
+**Additional context**
+Add any other context about the problem here.
+
+**Possible Solution**
+If you have an idea how to fix the issue, please describe it here.

.github/ISSUE_TEMPLATE/config.yml

@@ -0,0 +1,11 @@
+blank_issues_enabled: false
+contact_links:
+  - name: 📚 Documentation
+    url: https://github.com/PicPeak/picpeak/blob/main/DEPLOYMENT.md
+    about: Please read the documentation before opening an issue
+  - name: 💬 Discussions
+    url: https://github.com/PicPeak/picpeak/discussions
+    about: Ask questions and discuss with the community
+  - name: 🔒 Security Issues
+    url: https://github.com/PicPeak/picpeak/blob/main/SECURITY.md
+    about: Please review our security policy for reporting vulnerabilities

.github/ISSUE_TEMPLATE/documentation.md

@@ -0,0 +1,33 @@
+---
+name: Documentation
+about: Report issues or improvements needed in documentation
+title: '[DOCS] '
+labels: 'documentation'
+assignees: ''
+
+---
+
+**What documentation needs improvement?**
+Please specify which document or section needs attention:
+- [ ] README.md
+- [ ] DEPLOYMENT.md
+- [ ] CONTRIBUTING.md
+- [ ] API Documentation
+- [ ] Code Comments
+- [ ] Other: ___________
+
+**Describe the issue**
+What's wrong or missing in the documentation?
+
+**Suggested improvement**
+How would you improve this documentation?
+
+**Target audience**
+Who is this documentation for?
+- [ ] New users setting up PicPeak
+- [ ] Developers contributing to the project
+- [ ] System administrators
+- [ ] End users (photographers/clients)
+
+**Additional context**
+Add any other context, examples, or references here.

.github/ISSUE_TEMPLATE/feature_request.md

@@ -0,0 +1,38 @@
+---
+name: Feature request
+about: Suggest an idea for PicPeak
+title: '[FEATURE] '
+labels: 'enhancement'
+assignees: ''
+
+---
+
+**Is your feature request related to a problem? Please describe.**
+A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]
+
+**Describe the solution you'd like**
+A clear and concise description of what you want to happen.
+
+**Describe alternatives you've considered**
+A clear and concise description of any alternative solutions or features you've considered.
+
+**Use Case**
+Please describe how this feature would be used:
+- Who would use it? (photographers, clients, admins)
+- When would they use it?
+- Why is it important?
+
+**Similar Features**
+Are there similar features in:
+- PicDrop
+- Scrapbook.de
+- Other photo sharing platforms
+
+**Mockups or Examples**
+If applicable, add mockups, diagrams, or links to similar implementations.
+
+**Additional context**
+Add any other context or screenshots about the feature request here.
+
+**Implementation Ideas**
+If you have technical ideas about how this could be implemented, please share them.

.github/ISSUE_TEMPLATE/question.md

@@ -0,0 +1,26 @@
+---
+name: Question
+about: Ask a question about PicPeak
+title: '[QUESTION] '
+labels: 'question'
+assignees: ''
+
+---
+
+**Question**
+What would you like to know about PicPeak?
+
+**Context**
+Please provide context to help us answer your question better:
+- What are you trying to achieve?
+- What have you already tried?
+- Which documentation have you consulted?
+
+**Environment**
+If relevant to your question:
+- PicPeak Version:
+- Deployment Method:
+- Operating System:
+
+**Related Issues or Discussions**
+Link to any related issues, discussions, or documentation.

.github/ISSUE_TEMPLATE/security_vulnerability.md

@@ -0,0 +1,37 @@
+---
+name: Security Vulnerability
+about: Report security issues privately
+title: '[SECURITY] '
+labels: 'security'
+assignees: ''
+
+---
+
+⚠️ **IMPORTANT: For serious security vulnerabilities, please DO NOT create a public issue.**
+
+Instead, please use [GitHub Private Vulnerability Reporting](https://github.com/PicPeak/picpeak/security/advisories/new) or email **info@picpeak.app** with the details.
+
+For minor security improvements or questions, you can use this template:
+
+**Type of Security Issue**
+- [ ] Authentication/Authorization
+- [ ] Data Exposure
+- [ ] Input Validation
+- [ ] Configuration Issue
+- [ ] Dependency Vulnerability
+- [ ] Other: ___________
+
+**Description**
+Brief description of the security concern.
+
+**Impact**
+What could an attacker potentially do?
+
+**Steps to Reproduce**
+If applicable, how can this be reproduced?
+
+**Suggested Fix**
+If you have ideas on how to fix this issue.
+
+**References**
+Any relevant security advisories, CVEs, or documentation.

.github/pull_request_template.md

@@ -0,0 +1,49 @@
+## Description
+
+Please include a summary of the changes and which issue is fixed. Include relevant motivation and context.
+
+Fixes # (issue)
+
+## Type of change
+
+Please delete options that are not relevant.
+
+- [ ] Bug fix (non-breaking change which fixes an issue)
+- [ ] New feature (non-breaking change which adds functionality)
+- [ ] Breaking change (fix or feature that would cause existing functionality to not work as expected)
+- [ ] Documentation update
+- [ ] Performance improvement
+- [ ] Code refactoring
+
+## How Has This Been Tested?
+
+Please describe the tests that you ran to verify your changes. Provide instructions so we can reproduce.
+
+- [ ] Unit tests pass (`npm test`)
+- [ ] Manual testing completed
+- [ ] Tested on Docker deployment
+- [ ] Tested on production-like environment
+
+**Test Configuration**:
+* PicPeak Version:
+* Node.js Version:
+* Database: PostgreSQL / SQLite
+* Browser:
+
+## Checklist:
+
+- [ ] My code follows the style guidelines of this project
+- [ ] I have performed a self-review of my code
+- [ ] I have commented my code, particularly in hard-to-understand areas
+- [ ] I have made corresponding changes to the documentation
+- [ ] My changes generate no new warnings
+- [ ] I have added tests that prove my fix is effective or that my feature works
+- [ ] New and existing unit tests pass locally with my changes
+- [ ] Any dependent changes have been merged and published
+- [ ] I have updated the CHANGELOG.md file
+
+## Screenshots (if appropriate):
+
+## Additional Notes:
+
+Add any additional notes, concerns, or discussion points here.

.github/workflows/README-DOCKER.md

@@ -0,0 +1,213 @@
+# Docker Build and Push Workflow
+
+This GitHub Actions workflow automatically builds and pushes Docker images for both the backend and frontend to GitHub Container Registry (ghcr.io).
+
+## Features
+
+- 🔧 **Automatic builds** on push to main/develop branches, PRs, and releases
+- 🏗️ **Multi-architecture support** (linux/amd64 and linux/arm64)
+- 🏷️ **Smart tagging** based on branches, versions, and commits
+- 🔒 **Security scanning** with Trivy vulnerability scanner
+- 💾 **Build caching** for faster subsequent builds
+- 📊 **Build summaries** in GitHub Actions UI
+
+## Authentication
+
+The workflow uses the built-in `GITHUB_TOKEN` for authentication with GitHub Container Registry. No additional setup or personal access tokens are required.
+
+### Required Permissions
+
+The workflow automatically sets the necessary permissions:
+- `contents: read` - To checkout the repository
+- `packages: write` - To push images to ghcr.io
+- `security-events: write` - To upload security scan results
+
+## Image Tags
+
+Images are automatically tagged based on the trigger event:
+
+| Event | Tags Generated |
+|-------|---------------|
+| Push to main | `latest`, `main`, `main-<short-sha>` |
+| Push to develop | `develop`, `develop-<short-sha>` |
+| Pull Request | `pr-<number>` |
+| Release (v1.2.3) | `1.2.3`, `1.2`, `1`, `latest` |
+| Manual trigger | Based on branch + optional push |
+
+## Usage
+
+### Pull Images
+
+Once published, images can be pulled using:
+
+```bash
+# Pull backend image
+docker pull ghcr.io/picpeak/picpeak/backend:latest
+
+# Pull frontend image
+docker pull ghcr.io/picpeak/picpeak/frontend:latest
+
+# Pull specific version
+docker pull ghcr.io/picpeak/picpeak/backend:v1.0.0
+
+# Pull for specific architecture
+docker pull --platform linux/arm64 ghcr.io/picpeak/picpeak/backend:latest
+```
+
+### Using in Docker Compose
+
+```yaml
+version: '3.8'
+
+services:
+  backend:
+    image: ghcr.io/picpeak/picpeak/backend:latest
+    environment:
+      - NODE_ENV=production
+    ports:
+      - "3001:3000"
+
+  frontend:
+    image: ghcr.io/picpeak/picpeak/frontend:latest
+    ports:
+      - "80:80"
+```
+
+### Using in Kubernetes
+
+```yaml
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: picpeak-backend
+spec:
+  replicas: 3
+  template:
+    spec:
+      containers:
+      - name: backend
+        image: ghcr.io/picpeak/picpeak/backend:latest
+        imagePullPolicy: Always
+```
+
+## Manual Workflow Trigger
+
+You can manually trigger the workflow from the Actions tab:
+
+1. Go to Actions → "Build and Push Docker Images"
+2. Click "Run workflow"
+3. Select branch and whether to push images
+4. Click "Run workflow"
+
+## Security Scanning
+
+The workflow includes Trivy vulnerability scanning that:
+- Scans for CRITICAL and HIGH severity vulnerabilities
+- Uploads results to GitHub Security tab
+- Available under Security → Code scanning alerts
+
+## Build Optimization
+
+The workflow uses several optimization techniques:
+
+1. **GitHub Actions Cache**: Speeds up builds by caching layers
+2. **Multi-stage builds**: Reduces final image size
+3. **Parallel builds**: Backend and frontend build simultaneously
+4. **Smart rebuilds**: Only rebuilds changed components
+
+## Troubleshooting
+
+### Permission Denied Errors
+
+If you encounter permission errors when pushing images:
+
+1. **First-time setup**: The first push creates a private package. You may need to:
+   - Go to your package settings at `https://github.com/users/YOUR_USERNAME/packages`
+   - Link the package to your repository
+   - Set package visibility (public/private)
+
+2. **Organization repositories**: Ensure the organization allows GitHub Actions to create packages
+
+### Build Failures
+
+Check the workflow logs in the Actions tab for detailed error messages. Common issues:
+- Missing dependencies in package.json
+- Dockerfile syntax errors
+- Network issues during package installation
+
+### Image Not Found
+
+If images aren't visible after successful push:
+- Check package visibility settings
+- Ensure you're authenticated to pull private images:
+  ```bash
+  echo $GITHUB_TOKEN | docker login ghcr.io -u YOUR_USERNAME --password-stdin
+  ```
+
+## Package Management
+
+### View Packages
+
+Your Docker images are available at:
+- Backend: `https://github.com/orgs/PicPeak/packages/container/package/picpeak%2Fbackend`
+- Frontend: `https://github.com/orgs/PicPeak/packages/container/package/picpeak%2Ffrontend`
+
+### Delete Old Versions
+
+To save storage, you can delete old versions:
+1. Go to package settings
+2. Click on "Manage versions"
+3. Select versions to delete
+4. Click "Delete selected versions"
+
+### Set Retention Policy
+
+Configure automatic cleanup in package settings:
+1. Go to package settings
+2. Click on "Manage Actions access"
+3. Set retention days for untagged versions
+
+## Best Practices
+
+1. **Use semantic versioning** for releases (e.g., v1.2.3)
+2. **Test images locally** before pushing to production
+3. **Monitor security alerts** from Trivy scans
+4. **Clean up old images** regularly to save storage
+5. **Use specific tags** in production (avoid `latest`)
+
+## Advanced Configuration
+
+### Custom Registry
+
+To use a different registry, update the workflow:
+
+```yaml
+env:
+  REGISTRY: docker.io  # or your custom registry
+  BACKEND_IMAGE_NAME: yourusername/picpeak-backend
+```
+
+### Additional Platforms
+
+To build for more platforms:
+
+```yaml
+platforms: linux/amd64,linux/arm64,linux/arm/v7
+```
+
+### Custom Build Arguments
+
+Add build arguments in the workflow:
+
+```yaml
+build-args: |
+  NODE_VERSION=20
+  API_URL=${{ secrets.API_URL }}
+```
+
+## Related Documentation
+
+- [GitHub Container Registry Docs](https://docs.github.com/en/packages/working-with-a-github-packages-registry/working-with-the-container-registry)
+- [Docker Build Action](https://github.com/docker/build-push-action)
+- [Trivy Security Scanner](https://github.com/aquasecurity/trivy)
+- [Multi-platform Builds](https://docs.docker.com/build/building/multi-platform/)

.github/workflows/bypass-size-gate.yml

@@ -0,0 +1,70 @@
+name: Bypass size gate
+
+# Caps how large a PR a "review-bypass" collaborator (e.g. @Luca-Timo) can
+# self-merge without a maintainer review. The branch-protection bypass list
+# alone is binary — once a user is on it they can merge anything without
+# review. This workflow reports a REQUIRED status check that fails when a
+# bypass user's PR exceeds the configured size threshold, which blocks the
+# merge even with bypass enabled. Other contributors are unaffected (the
+# check reports success for them so the required-check gate doesn't trip).
+#
+# To tune: edit LINE_LIMIT or BYPASS_USERS below.
+#
+# Trigger note: uses `pull_request_target` so the workflow has the elevated
+# permissions of the base repo's GITHUB_TOKEN (read PR metadata, write
+# checks). The script never executes code FROM the PR — it only reads
+# metadata via the API — so this is safe against fork-PR attacks.
+
+on:
+  pull_request_target:
+    types: [opened, reopened, synchronize, ready_for_review]
+
+permissions:
+  pull-requests: read
+  checks: write
+
+jobs:
+  size-gate:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Compute PR size and report check status
+        uses: actions/github-script@v7
+        with:
+          script: |
+            // Tune these two constants if the policy shifts.
+            const LINE_LIMIT = 300;
+            const BYPASS_USERS = ['Luca-Timo'];
+
+            const pr = context.payload.pull_request;
+            const author = pr.user.login;
+            const linesChanged = pr.additions + pr.deletions;
+            const filesChanged = pr.changed_files;
+
+            let conclusion, title, summary;
+
+            if (!BYPASS_USERS.includes(author)) {
+              // Not a bypass user — this gate doesn't apply to them. They
+              // go through normal review. Report success so the required
+              // check doesn't block their merge.
+              conclusion = 'success';
+              title = 'Not applicable';
+              summary = `This gate only restricts review-bypass for: ${BYPASS_USERS.join(', ')}. PRs from other authors (${author} here) go through the normal review path and are unaffected.`;
+            } else if (linesChanged <= LINE_LIMIT) {
+              conclusion = 'success';
+              title = `OK — within bypass limit (${linesChanged} lines)`;
+              summary = `Small PR: ${linesChanged} lines changed across ${filesChanged} file(s). Within the ${LINE_LIMIT}-line self-merge limit for @${author}. Can be merged without a maintainer review.`;
+            } else {
+              conclusion = 'failure';
+              title = `Too large for bypass (${linesChanged} lines)`;
+              summary = `Large PR: ${linesChanged} lines changed across ${filesChanged} file(s). Exceeds the ${LINE_LIMIT}-line self-merge limit for @${author} — needs an approving review from a maintainer before merge. Split into smaller PRs or wait for review.`;
+            }
+
+            await github.rest.checks.create({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              name: 'bypass-size-gate',
+              head_sha: pr.head.sha,
+              status: 'completed',
+              conclusion,
+              output: { title, summary }
+            });

.github/workflows/docker-build.yml

@@ -0,0 +1,603 @@
+name: Build and Push Docker Images
+
+# This workflow is triggered by:
+# - Push to main/stable branches (main → ':main' rolling tag for active-dev
+#   builds; stable → ':stable' + ':latest' for the curated channel)
+# - Version tags from Release Please (e.g., v1.2.0 -> builds versioned images)
+# - GitHub Releases (created by Release Please)
+# - Pull requests (build verification only, no push by default)
+# - Manual workflow dispatch
+#
+# Multi-arch strategy:
+#   Each image (backend, frontend) is built once per architecture on a
+#   native runner — linux/amd64 on ubuntu-latest, linux/arm64 on
+#   ubuntu-24.04-arm. Each leg pushes by digest to GHCR. A follow-up
+#   merge job combines the digests into a multi-arch manifest and applies
+#   the human-readable tags. This is the pattern documented at
+#   https://docs.docker.com/build/ci/github-actions/multi-platform/
+#
+#   Native runners are used instead of QEMU because npm install under
+#   QEMU was previously too slow/unreliable for regular branch builds.
+
+on:
+  push:
+    branches: [ main, stable ]
+    tags: [ 'v*.*.*', 'v*.*.*-beta.*' ]  # Triggered by Release Please tags (stable and beta)
+  pull_request:
+    branches: [ main, stable ]
+  release:
+    types: [ published ]  # Triggered when Release Please creates a release
+  workflow_dispatch:
+    inputs:
+      push:
+        description: 'Push images to registry'
+        required: false
+        default: 'false'
+        type: choice
+        options:
+          - 'true'
+          - 'false'
+
+# Once release-please authors releases with a PAT (#719), a new version fires
+# BOTH the tag-push and the release-published triggers (GITHUB_TOKEN used to
+# suppress them). They build the same immutable version, so collapse them into a
+# single run by grouping on the ref. Branch and PR builds use different refs and
+# still run independently; a superseding push cancels an in-flight run for the
+# same ref (only the newest build per ref is kept).
+concurrency:
+  group: docker-build-${{ github.ref }}
+  cancel-in-progress: true
+
+env:
+  REGISTRY: ghcr.io
+  # BACKEND_IMAGE_NAME and FRONTEND_IMAGE_NAME are computed per job in the
+  # "Compute image names" step. GHCR requires all-lowercase repository names,
+  # but ${{ github.repository }} preserves the original case (e.g. "Luca-Timo/...").
+  # Computing them with bash parameter expansion (${VAR,,}) keeps the workflow
+  # working on forks regardless of the owner's name casing.
+
+# Default GITHUB_TOKEN to read-only at the workflow level. Each job that
+# needs to publish to GHCR sets `packages: write` explicitly. This keeps
+# the rest of the workflow (and any future steps) from inheriting unneeded
+# privileges (CKV2_GHA_1).
+permissions:
+  contents: read
+
+jobs:
+  # -----------------------------------------------------------------------------
+  # Backend: per-arch build, then merge into a multi-arch manifest
+  # -----------------------------------------------------------------------------
+  build-backend:
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - platform: linux/amd64
+            runner: ubuntu-latest
+          - platform: linux/arm64
+            runner: ubuntu-24.04-arm
+    runs-on: ${{ matrix.runner }}
+    permissions:
+      contents: read
+      packages: write
+      # Trivy uploads its SARIF to the Security tab from this job — see
+      # the "Run Trivy" step below. Scanning per-arch by digest (#476)
+      # is reliable; scanning the multi-arch index by tag from the
+      # merge-* job was not.
+      security-events: write
+
+    steps:
+    - name: Checkout code
+      uses: actions/checkout@v4
+
+    - name: Compute image names (lowercase for GHCR)
+      run: |
+        repo_lc="${GITHUB_REPOSITORY,,}"
+        echo "BACKEND_IMAGE_NAME=${repo_lc}/backend" >> "$GITHUB_ENV"
+        echo "FRONTEND_IMAGE_NAME=${repo_lc}/frontend" >> "$GITHUB_ENV"
+
+    - name: Prepare platform pair
+      run: |
+        platform="${{ matrix.platform }}"
+        echo "PLATFORM_PAIR=${platform//\//-}" >> "$GITHUB_ENV"
+
+    - name: Set up Docker Buildx
+      uses: docker/setup-buildx-action@v3
+
+    - name: Log in to Container Registry
+      if: github.event_name != 'pull_request' || github.event.inputs.push == 'true'
+      id: login-ghcr
+      continue-on-error: true
+      uses: docker/login-action@v3
+      with:
+        registry: ${{ env.REGISTRY }}
+        username: ${{ github.actor }}
+        password: ${{ secrets.GITHUB_TOKEN }}
+
+    - name: Determine if pushing
+      id: push-decision
+      run: |
+        if [[ "${{ github.event_name }}" == "pull_request" && "${{ github.event.inputs.push }}" != "true" ]]; then
+          echo "push=false" >> "$GITHUB_OUTPUT"
+        elif [[ "${{ steps.login-ghcr.outcome }}" != "success" ]]; then
+          echo "push=false" >> "$GITHUB_OUTPUT"
+        else
+          echo "push=true" >> "$GITHUB_OUTPUT"
+        fi
+
+    - name: Extract metadata for Backend (labels only)
+      id: meta-backend
+      uses: docker/metadata-action@v5
+      with:
+        images: ${{ env.REGISTRY }}/${{ env.BACKEND_IMAGE_NAME }}
+        labels: |
+          org.opencontainers.image.title=PicPeak Backend
+          org.opencontainers.image.description=PicPeak photo sharing platform backend service
+          org.opencontainers.image.vendor=PicPeak
+          maintainer=${{ github.repository_owner }}
+
+    - name: Build Backend image (push by digest)
+      id: build
+      uses: docker/build-push-action@v5
+      with:
+        context: ./backend
+        file: ./backend/Dockerfile
+        platforms: ${{ matrix.platform }}
+        labels: ${{ steps.meta-backend.outputs.labels }}
+        cache-from: type=gha,scope=backend-${{ env.PLATFORM_PAIR }}
+        # ignore-error: a flaky GitHub Actions cache write ("error writing
+        # layer blob: not_found") must not fail an otherwise-successful build
+        # that already pushed the image.
+        cache-to: type=gha,mode=max,scope=backend-${{ env.PLATFORM_PAIR }},ignore-error=true
+        outputs: ${{ steps.push-decision.outputs.push == 'true' && format('type=image,name={0}/{1},push-by-digest=true,name-canonical=true,push=true', env.REGISTRY, env.BACKEND_IMAGE_NAME) || 'type=cacheonly' }}
+        build-args: |
+          CACHEBUST=${{ github.run_number }}
+          BUILD_DATE=${{ github.event.head_commit.timestamp }}
+          VCS_REF=${{ github.sha }}
+          VERSION=${{ steps.meta-backend.outputs.version }}
+
+    - name: Export digest
+      if: steps.push-decision.outputs.push == 'true'
+      run: |
+        mkdir -p /tmp/digests
+        digest="${{ steps.build.outputs.digest }}"
+        touch "/tmp/digests/${digest#sha256:}"
+
+    - name: Upload digest artifact
+      if: steps.push-decision.outputs.push == 'true'
+      uses: actions/upload-artifact@v4
+      with:
+        name: digests-backend-${{ env.PLATFORM_PAIR }}
+        path: /tmp/digests/*
+        if-no-files-found: error
+        retention-days: 1
+
+    # Per-arch vulnerability scan (#476). Scanning the multi-arch
+    # manifest from the merge-* job by tag is unreliable — Trivy's
+    # remote resolver crashes intermittently with "no child with
+    # platform linux/amd64 in index". The fix is to scan each leg
+    # by its single-platform digest right here, where it just landed
+    # in GHCR. Tag pinned (was @master) so the action + bundled
+    # Trivy binary don't float between runs.
+    #
+    # exit-code is left unset (=0) for now: Trivy reports findings
+    # to the Security tab but doesn't fail the build. Flipping that
+    # to '1' to actually gate CI is a deliberate follow-up — needs an
+    # audit pass first so the next beta build doesn't surprise red.
+    - name: Run Trivy vulnerability scanner (per-arch, by digest)
+      if: steps.push-decision.outputs.push == 'true'
+      uses: aquasecurity/trivy-action@v0.36.0
+      env:
+        # docker/build-push-action wraps every push in an OCI index
+        # (carries the SLSA provenance attestation alongside the
+        # actual image). Trivy's remote backend defaults to
+        # linux/amd64 regardless of host arch when resolving an
+        # index, which makes the arm64 leg crash with "no child
+        # with platform linux/amd64". Telling Trivy which child to
+        # scan keeps the provenance attestation intact and fixes
+        # the resolver crash. Pin to matrix.platform so each leg
+        # scans its own arch.
+        TRIVY_PLATFORM: ${{ matrix.platform }}
+      with:
+        image-ref: ${{ env.REGISTRY }}/${{ env.BACKEND_IMAGE_NAME }}@${{ steps.build.outputs.digest }}
+        format: 'sarif'
+        output: 'trivy-backend-${{ env.PLATFORM_PAIR }}.sarif'
+        severity: 'CRITICAL,HIGH'
+        # Base-image CVEs with no released fix are not actionable: the
+        # Dockerfiles already run `apt-get upgrade -y` behind a CACHEBUST,
+        # so a fix lands in the next build automatically. Reporting them
+        # buries the findings someone can actually do something about.
+        # Dropping them is also the precondition for ever setting
+        # exit-code: 1, which build-backend's comment flags as a
+        # deliberate follow-up.
+        ignore-unfixed: true
+        timeout: '10m'
+
+    - name: Upload Trivy scan results to GitHub Security tab
+      if: steps.push-decision.outputs.push == 'true'
+      uses: github/codeql-action/upload-sarif@v4
+      with:
+        sarif_file: 'trivy-backend-${{ env.PLATFORM_PAIR }}.sarif'
+        # Distinct category per arch so the Security tab surfaces
+        # per-platform findings independently — an amd64-only CVE in
+        # a base layer doesn't get masked by the arm64 scan.
+        category: 'backend-vulnerabilities-${{ env.PLATFORM_PAIR }}'
+
+  merge-backend:
+    needs: build-backend
+    runs-on: ubuntu-latest
+    # No security-events permission here — vulnerability scanning moved
+    # to per-arch build-backend jobs (#476). This job's only job is to
+    # combine the per-arch digests into a multi-arch manifest.
+    permissions:
+      contents: read
+      packages: write
+    # Only run when at least one digest was pushed (i.e. not on PRs without push intent).
+    if: github.event_name != 'pull_request' || github.event.inputs.push == 'true'
+
+    steps:
+    - name: Compute image names (lowercase for GHCR)
+      run: |
+        repo_lc="${GITHUB_REPOSITORY,,}"
+        echo "BACKEND_IMAGE_NAME=${repo_lc}/backend" >> "$GITHUB_ENV"
+        echo "FRONTEND_IMAGE_NAME=${repo_lc}/frontend" >> "$GITHUB_ENV"
+
+    - name: Download digest artifacts
+      uses: actions/download-artifact@v4
+      with:
+        path: /tmp/digests
+        pattern: digests-backend-*
+        merge-multiple: true
+
+    - name: Set up Docker Buildx
+      uses: docker/setup-buildx-action@v3
+
+    - name: Log in to Container Registry
+      id: login-ghcr
+      continue-on-error: true
+      uses: docker/login-action@v3
+      with:
+        registry: ${{ env.REGISTRY }}
+        username: ${{ github.actor }}
+        password: ${{ secrets.GITHUB_TOKEN }}
+
+    - name: Determine build context
+      id: context
+      run: |
+        if [[ "${{ github.ref }}" == refs/tags/v*-beta* ]] || [[ "${{ github.ref }}" == refs/heads/main ]]; then
+          # Active-dev branch (`main`, renamed from `beta` per #669) produces
+          # prereleases; the `-beta.N` version-suffix scheme is unchanged.
+          echo "channel=beta" >> $GITHUB_OUTPUT
+          echo "is_prerelease=true" >> $GITHUB_OUTPUT
+        else
+          echo "channel=stable" >> $GITHUB_OUTPUT
+          echo "is_prerelease=false" >> $GITHUB_OUTPUT
+        fi
+
+    - name: Extract metadata for Backend
+      id: meta-backend
+      uses: docker/metadata-action@v5
+      with:
+        images: ${{ env.REGISTRY }}/${{ env.BACKEND_IMAGE_NAME }}
+        labels: |
+          org.opencontainers.image.title=PicPeak Backend
+          org.opencontainers.image.description=PicPeak photo sharing platform backend service
+          org.opencontainers.image.vendor=PicPeak
+          maintainer=${{ github.repository_owner }}
+        tags: |
+          type=ref,event=branch
+          type=ref,event=pr
+          type=semver,pattern={{version}}
+          type=semver,pattern={{major}}.{{minor}},enable=${{ steps.context.outputs.is_prerelease == 'false' }}
+          type=semver,pattern={{major}},enable=${{ steps.context.outputs.is_prerelease == 'false' }}
+          type=sha,format=short
+          # `:latest` + `:stable` follow the stable channel (the `stable` branch +
+          # stable release tags). The default branch is now `main` (active dev),
+          # so `is_default_branch` no longer maps to "stable" — be explicit.
+          type=raw,value=latest,enable=${{ github.ref == 'refs/heads/stable' || (startsWith(github.ref, 'refs/tags/v') && steps.context.outputs.is_prerelease == 'false') }}
+          type=raw,value=stable,enable=${{ github.ref == 'refs/heads/stable' || (startsWith(github.ref, 'refs/tags/v') && steps.context.outputs.is_prerelease == 'false') }}
+          # `:beta` is RETIRED post-rename (Option B / #669). Active-dev pulls
+          # are `:main` (auto via type=ref,event=branch). The pre-rename `:beta`
+          # tag remains frozen at its last build — operators should update.
+
+    - name: Create and push multi-arch manifest
+      working-directory: /tmp/digests
+      run: |
+        docker buildx imagetools create $(jq -cr '.tags | map("-t " + .) | join(" ")' <<< "$DOCKER_METADATA_OUTPUT_JSON") \
+          $(printf "${{ env.REGISTRY }}/${{ env.BACKEND_IMAGE_NAME }}@sha256:%s " *)
+
+    - name: Inspect manifest
+      run: |
+        docker buildx imagetools inspect ${{ env.REGISTRY }}/${{ env.BACKEND_IMAGE_NAME }}:${{ steps.meta-backend.outputs.version }}
+
+  # -----------------------------------------------------------------------------
+  # Frontend: per-arch build, then merge into a multi-arch manifest
+  # -----------------------------------------------------------------------------
+  build-frontend:
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - platform: linux/amd64
+            runner: ubuntu-latest
+          - platform: linux/arm64
+            runner: ubuntu-24.04-arm
+    runs-on: ${{ matrix.runner }}
+    permissions:
+      contents: read
+      packages: write
+      # See build-backend for the rationale (#476). Same pattern: per-arch
+      # vulnerability scan by digest, SARIF uploaded to the Security tab.
+      security-events: write
+
+    steps:
+    - name: Checkout code
+      uses: actions/checkout@v4
+
+    - name: Compute image names (lowercase for GHCR)
+      run: |
+        repo_lc="${GITHUB_REPOSITORY,,}"
+        echo "BACKEND_IMAGE_NAME=${repo_lc}/backend" >> "$GITHUB_ENV"
+        echo "FRONTEND_IMAGE_NAME=${repo_lc}/frontend" >> "$GITHUB_ENV"
+
+    - name: Prepare platform pair
+      run: |
+        platform="${{ matrix.platform }}"
+        echo "PLATFORM_PAIR=${platform//\//-}" >> "$GITHUB_ENV"
+
+    - name: Set up Docker Buildx
+      uses: docker/setup-buildx-action@v3
+
+    - name: Log in to Container Registry
+      if: github.event_name != 'pull_request' || github.event.inputs.push == 'true'
+      id: login-ghcr
+      continue-on-error: true
+      uses: docker/login-action@v3
+      with:
+        registry: ${{ env.REGISTRY }}
+        username: ${{ github.actor }}
+        password: ${{ secrets.GITHUB_TOKEN }}
+
+    - name: Determine if pushing
+      id: push-decision
+      run: |
+        if [[ "${{ github.event_name }}" == "pull_request" && "${{ github.event.inputs.push }}" != "true" ]]; then
+          echo "push=false" >> "$GITHUB_OUTPUT"
+        elif [[ "${{ steps.login-ghcr.outcome }}" != "success" ]]; then
+          echo "push=false" >> "$GITHUB_OUTPUT"
+        else
+          echo "push=true" >> "$GITHUB_OUTPUT"
+        fi
+
+    - name: Extract metadata for Frontend (labels only)
+      id: meta-frontend
+      uses: docker/metadata-action@v5
+      with:
+        images: ${{ env.REGISTRY }}/${{ env.FRONTEND_IMAGE_NAME }}
+        labels: |
+          org.opencontainers.image.title=PicPeak Frontend
+          org.opencontainers.image.description=PicPeak photo sharing platform frontend application
+          org.opencontainers.image.vendor=PicPeak
+          maintainer=${{ github.repository_owner }}
+
+    - name: Build Frontend image (push by digest)
+      id: build
+      uses: docker/build-push-action@v5
+      with:
+        context: ./frontend
+        file: ./frontend/Dockerfile
+        platforms: ${{ matrix.platform }}
+        labels: ${{ steps.meta-frontend.outputs.labels }}
+        cache-from: type=gha,scope=frontend-${{ env.PLATFORM_PAIR }}
+        # ignore-error: a flaky GitHub Actions cache write ("error writing
+        # layer blob: not_found") must not fail an otherwise-successful build
+        # that already pushed the image.
+        cache-to: type=gha,mode=max,scope=frontend-${{ env.PLATFORM_PAIR }},ignore-error=true
+        outputs: ${{ steps.push-decision.outputs.push == 'true' && format('type=image,name={0}/{1},push-by-digest=true,name-canonical=true,push=true', env.REGISTRY, env.FRONTEND_IMAGE_NAME) || 'type=cacheonly' }}
+        build-args: |
+          CACHEBUST=${{ github.run_number }}
+          BUILD_DATE=${{ github.event.head_commit.timestamp }}
+          VCS_REF=${{ github.sha }}
+          VERSION=${{ steps.meta-frontend.outputs.version }}
+
+    - name: Export digest
+      if: steps.push-decision.outputs.push == 'true'
+      run: |
+        mkdir -p /tmp/digests
+        digest="${{ steps.build.outputs.digest }}"
+        touch "/tmp/digests/${digest#sha256:}"
+
+    - name: Upload digest artifact
+      if: steps.push-decision.outputs.push == 'true'
+      uses: actions/upload-artifact@v4
+      with:
+        name: digests-frontend-${{ env.PLATFORM_PAIR }}
+        path: /tmp/digests/*
+        if-no-files-found: error
+        retention-days: 1
+
+    # Per-arch vulnerability scan (#476). See build-backend for the
+    # full rationale; identical pattern here, only the image-ref +
+    # SARIF filename + category change.
+    - name: Run Trivy vulnerability scanner (per-arch, by digest)
+      if: steps.push-decision.outputs.push == 'true'
+      uses: aquasecurity/trivy-action@v0.36.0
+      env:
+        # See build-backend for the rationale — pin Trivy's platform
+        # to the matrix arch so its remote-index resolver picks the
+        # right child instead of defaulting to linux/amd64 and
+        # crashing on the arm64 leg.
+        TRIVY_PLATFORM: ${{ matrix.platform }}
+      with:
+        image-ref: ${{ env.REGISTRY }}/${{ env.FRONTEND_IMAGE_NAME }}@${{ steps.build.outputs.digest }}
+        format: 'sarif'
+        output: 'trivy-frontend-${{ env.PLATFORM_PAIR }}.sarif'
+        severity: 'CRITICAL,HIGH'
+        # Base-image CVEs with no released fix are not actionable: the
+        # Dockerfiles already run `apt-get upgrade -y` behind a CACHEBUST,
+        # so a fix lands in the next build automatically. Reporting them
+        # buries the findings someone can actually do something about.
+        # Dropping them is also the precondition for ever setting
+        # exit-code: 1, which build-backend's comment flags as a
+        # deliberate follow-up.
+        ignore-unfixed: true
+        timeout: '10m'
+
+    - name: Upload Trivy scan results to GitHub Security tab
+      if: steps.push-decision.outputs.push == 'true'
+      uses: github/codeql-action/upload-sarif@v4
+      with:
+        sarif_file: 'trivy-frontend-${{ env.PLATFORM_PAIR }}.sarif'
+        category: 'frontend-vulnerabilities-${{ env.PLATFORM_PAIR }}'
+
+  merge-frontend:
+    needs: build-frontend
+    runs-on: ubuntu-latest
+    # See merge-backend — vulnerability scanning moved to the per-arch
+    # build-frontend matrix (#476). This job only publishes the manifest.
+    permissions:
+      contents: read
+      packages: write
+    if: github.event_name != 'pull_request' || github.event.inputs.push == 'true'
+
+    steps:
+    - name: Compute image names (lowercase for GHCR)
+      run: |
+        repo_lc="${GITHUB_REPOSITORY,,}"
+        echo "BACKEND_IMAGE_NAME=${repo_lc}/backend" >> "$GITHUB_ENV"
+        echo "FRONTEND_IMAGE_NAME=${repo_lc}/frontend" >> "$GITHUB_ENV"
+
+    - name: Download digest artifacts
+      uses: actions/download-artifact@v4
+      with:
+        path: /tmp/digests
+        pattern: digests-frontend-*
+        merge-multiple: true
+
+    - name: Set up Docker Buildx
+      uses: docker/setup-buildx-action@v3
+
+    - name: Log in to Container Registry
+      id: login-ghcr
+      continue-on-error: true
+      uses: docker/login-action@v3
+      with:
+        registry: ${{ env.REGISTRY }}
+        username: ${{ github.actor }}
+        password: ${{ secrets.GITHUB_TOKEN }}
+
+    - name: Determine build context
+      id: context
+      run: |
+        if [[ "${{ github.ref }}" == refs/tags/v*-beta* ]] || [[ "${{ github.ref }}" == refs/heads/main ]]; then
+          # Active-dev branch (`main`, renamed from `beta` per #669) produces
+          # prereleases; the `-beta.N` version-suffix scheme is unchanged.
+          echo "channel=beta" >> $GITHUB_OUTPUT
+          echo "is_prerelease=true" >> $GITHUB_OUTPUT
+        else
+          echo "channel=stable" >> $GITHUB_OUTPUT
+          echo "is_prerelease=false" >> $GITHUB_OUTPUT
+        fi
+
+    - name: Extract metadata for Frontend
+      id: meta-frontend
+      uses: docker/metadata-action@v5
+      with:
+        images: ${{ env.REGISTRY }}/${{ env.FRONTEND_IMAGE_NAME }}
+        labels: |
+          org.opencontainers.image.title=PicPeak Frontend
+          org.opencontainers.image.description=PicPeak photo sharing platform frontend application
+          org.opencontainers.image.vendor=PicPeak
+          maintainer=${{ github.repository_owner }}
+        tags: |
+          type=ref,event=branch
+          type=ref,event=pr
+          type=semver,pattern={{version}}
+          type=semver,pattern={{major}}.{{minor}},enable=${{ steps.context.outputs.is_prerelease == 'false' }}
+          type=semver,pattern={{major}},enable=${{ steps.context.outputs.is_prerelease == 'false' }}
+          type=sha,format=short
+          # `:latest` + `:stable` follow the stable channel (the `stable` branch +
+          # stable release tags). The default branch is now `main` (active dev),
+          # so `is_default_branch` no longer maps to "stable" — be explicit.
+          type=raw,value=latest,enable=${{ github.ref == 'refs/heads/stable' || (startsWith(github.ref, 'refs/tags/v') && steps.context.outputs.is_prerelease == 'false') }}
+          type=raw,value=stable,enable=${{ github.ref == 'refs/heads/stable' || (startsWith(github.ref, 'refs/tags/v') && steps.context.outputs.is_prerelease == 'false') }}
+          # `:beta` is RETIRED post-rename (Option B / #669). Active-dev pulls
+          # are `:main` (auto via type=ref,event=branch). The pre-rename `:beta`
+          # tag remains frozen at its last build — operators should update.
+
+    - name: Create and push multi-arch manifest
+      working-directory: /tmp/digests
+      run: |
+        docker buildx imagetools create $(jq -cr '.tags | map("-t " + .) | join(" ")' <<< "$DOCKER_METADATA_OUTPUT_JSON") \
+          $(printf "${{ env.REGISTRY }}/${{ env.FRONTEND_IMAGE_NAME }}@sha256:%s " *)
+
+    - name: Inspect manifest
+      run: |
+        docker buildx imagetools inspect ${{ env.REGISTRY }}/${{ env.FRONTEND_IMAGE_NAME }}:${{ steps.meta-frontend.outputs.version }}
+
+  summary:
+    needs: [build-backend, merge-backend, build-frontend, merge-frontend]
+    if: always()
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+
+    steps:
+    - name: Compute image names (lowercase for GHCR)
+      run: |
+        repo_lc="${GITHUB_REPOSITORY,,}"
+        echo "BACKEND_IMAGE_NAME=${repo_lc}/backend" >> "$GITHUB_ENV"
+        echo "FRONTEND_IMAGE_NAME=${repo_lc}/frontend" >> "$GITHUB_ENV"
+
+    - name: Build Summary
+      run: |
+        echo "## 🐳 Docker Build Summary" >> $GITHUB_STEP_SUMMARY
+        echo "" >> $GITHUB_STEP_SUMMARY
+
+        if [[ "${{ needs.build-backend.result }}" == "success" ]]; then
+          echo "✅ **Backend build (per-arch)**: Successfully built" >> $GITHUB_STEP_SUMMARY
+        else
+          echo "❌ **Backend build (per-arch)**: ${{ needs.build-backend.result }}" >> $GITHUB_STEP_SUMMARY
+        fi
+
+        if [[ "${{ needs.merge-backend.result }}" == "success" ]]; then
+          echo "✅ **Backend manifest merge**: Successfully published" >> $GITHUB_STEP_SUMMARY
+        elif [[ "${{ needs.merge-backend.result }}" == "skipped" ]]; then
+          echo "ℹ️ **Backend manifest merge**: Skipped (verify-only build)" >> $GITHUB_STEP_SUMMARY
+        else
+          echo "❌ **Backend manifest merge**: ${{ needs.merge-backend.result }}" >> $GITHUB_STEP_SUMMARY
+        fi
+
+        if [[ "${{ needs.build-frontend.result }}" == "success" ]]; then
+          echo "✅ **Frontend build (per-arch)**: Successfully built" >> $GITHUB_STEP_SUMMARY
+        else
+          echo "❌ **Frontend build (per-arch)**: ${{ needs.build-frontend.result }}" >> $GITHUB_STEP_SUMMARY
+        fi
+
+        if [[ "${{ needs.merge-frontend.result }}" == "success" ]]; then
+          echo "✅ **Frontend manifest merge**: Successfully published" >> $GITHUB_STEP_SUMMARY
+        elif [[ "${{ needs.merge-frontend.result }}" == "skipped" ]]; then
+          echo "ℹ️ **Frontend manifest merge**: Skipped (verify-only build)" >> $GITHUB_STEP_SUMMARY
+        else
+          echo "❌ **Frontend manifest merge**: ${{ needs.merge-frontend.result }}" >> $GITHUB_STEP_SUMMARY
+        fi
+
+        echo "" >> $GITHUB_STEP_SUMMARY
+        echo "### 📦 Images" >> $GITHUB_STEP_SUMMARY
+        echo "- Backend: \`${{ env.REGISTRY }}/${{ env.BACKEND_IMAGE_NAME }}\`" >> $GITHUB_STEP_SUMMARY
+        echo "- Frontend: \`${{ env.REGISTRY }}/${{ env.FRONTEND_IMAGE_NAME }}\`" >> $GITHUB_STEP_SUMMARY
+
+        echo "" >> $GITHUB_STEP_SUMMARY
+        echo "### 🏗️ Architectures" >> $GITHUB_STEP_SUMMARY
+        echo "Published manifests include both \`linux/amd64\` and \`linux/arm64\` (built natively, no QEMU)." >> $GITHUB_STEP_SUMMARY
+
+        echo "" >> $GITHUB_STEP_SUMMARY
+        echo "### 🏷️ Tags" >> $GITHUB_STEP_SUMMARY
+        echo "Images are tagged based on:" >> $GITHUB_STEP_SUMMARY
+        echo "- Branch name (for branch pushes)" >> $GITHUB_STEP_SUMMARY
+        echo "- PR number (for pull requests, when push is enabled)" >> $GITHUB_STEP_SUMMARY
+        echo "- Version tags (for releases)" >> $GITHUB_STEP_SUMMARY
+        echo "- Short SHA" >> $GITHUB_STEP_SUMMARY
+        echo "- \`latest\` (for main branch)" >> $GITHUB_STEP_SUMMARY
+        echo "- \`stable\` (for main branch and stable releases)" >> $GITHUB_STEP_SUMMARY
+        echo "- \`beta\` (for beta branch and pre-releases)" >> $GITHUB_STEP_SUMMARY

.github/workflows/install-smoke.yml

@@ -0,0 +1,221 @@
+name: Fresh-install smoke
+
+# Verifies that a clean Postgres install boots cleanly under the same
+# conditions a new user hits on their first `docker compose up -d`. The
+# specific scenarios this guards against — see #484 for the original
+# reproduction:
+#
+#   1. Bind-mounted host directories owned by a UID other than 1001
+#      (the container's nodejs user). The entrypoint must self-chown
+#      and drop privileges via su-exec.
+#   2. Cold-start Postgres with no prior schema (the FK-order bug fixed
+#      in #494, the index/created_at error fixed in #511, and any
+#      future migration-order issue that only surfaces on an empty DB).
+#
+# Triggers only on changes that touch the install path so unrelated PRs
+# don't pay the build cost.
+
+on:
+  # No `paths:` filter — branch protection on `main` + `stable` lists
+  # `fresh-install` as a REQUIRED check, and a path-filtered trigger
+  # that skipped on unrelated PRs (e.g. frontend-only) would leave the
+  # required check "missing" forever and block the merge. Better to
+  # pay the boot cost on every PR than maintain a per-path allowlist
+  # that drifts as the install surface evolves. (Branches also updated
+  # post-#669 rename: beta → main, old main → stable.)
+  push:
+    branches: [main, stable]
+  pull_request:
+    branches: [main, stable]
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+jobs:
+  fresh-install:
+    runs-on: ubuntu-latest
+    timeout-minutes: 15
+
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v4
+
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v3
+
+      # Build for the runner's arch only — we just need a runnable image.
+      # The full multi-arch build is the docker-build workflow's job.
+      - name: Build backend image
+        uses: docker/build-push-action@v5
+        with:
+          context: ./backend
+          file: ./backend/Dockerfile
+          load: true
+          tags: picpeak-backend:smoke
+          cache-from: type=gha,scope=install-smoke
+          # ignore-error: a flaky GHA cache write must not fail the build.
+          cache-to: type=gha,mode=max,scope=install-smoke,ignore-error=true
+
+      - name: Create Docker network
+        run: docker network create picpeak-smoke
+
+      # Mount as UID 1000 (the typical GitHub Actions runner user, and a
+      # common mismatch case on Linux hosts). The entrypoint must chown
+      # this to 1001 itself — that's the regression we're guarding.
+      - name: Prepare host bind-mount dirs owned by UID 1000
+        run: |
+          mkdir -p smoke-mounts/storage smoke-mounts/data smoke-mounts/logs
+          chmod 755 smoke-mounts smoke-mounts/*
+          ls -ld smoke-mounts/*
+
+      - name: Start Postgres
+        run: |
+          docker run -d --name picpeak-smoke-pg --network picpeak-smoke \
+            -e POSTGRES_USER=picpeak \
+            -e POSTGRES_PASSWORD=smokepass \
+            -e POSTGRES_DB=picpeak_prod \
+            --health-cmd="pg_isready -U picpeak -d picpeak_prod" \
+            --health-interval=2s --health-timeout=2s --health-retries=30 \
+            postgres:15-alpine
+
+      - name: Wait for Postgres healthy
+        run: |
+          for i in $(seq 1 60); do
+            status=$(docker inspect -f '{{.State.Health.Status}}' picpeak-smoke-pg 2>/dev/null || echo starting)
+            if [ "$status" = "healthy" ]; then
+              echo "postgres healthy after ${i}s"
+              exit 0
+            fi
+            sleep 1
+          done
+          echo "postgres did not become healthy in 60s"
+          docker logs picpeak-smoke-pg
+          exit 1
+
+      - name: Start backend with mismatched-UID bind mounts (fresh install)
+        run: |
+          docker run -d --name picpeak-smoke-bk --network picpeak-smoke \
+            -e NODE_ENV=production \
+            -e JWT_SECRET=smoketestsecretvalueof32characters \
+            -e DB_HOST=picpeak-smoke-pg \
+            -e DB_USER=picpeak \
+            -e DB_PASSWORD=smokepass \
+            -e DB_NAME=picpeak_prod \
+            -e ADMIN_EMAIL=admin@smoke.local \
+            -e ADMIN_PASSWORD=smokeAdminPass12345 \
+            -e STORAGE_PATH=/app/storage \
+            -v "$PWD/smoke-mounts/storage:/app/storage" \
+            -v "$PWD/smoke-mounts/data:/app/data" \
+            -v "$PWD/smoke-mounts/logs:/app/logs" \
+            picpeak-backend:smoke
+
+      - name: Wait for backend healthy
+        run: |
+          for i in $(seq 1 120); do
+            status=$(docker inspect -f '{{.State.Status}}' picpeak-smoke-bk 2>/dev/null || echo missing)
+            health=$(docker inspect -f '{{.State.Health.Status}}' picpeak-smoke-bk 2>/dev/null || echo none)
+            if [ "$status" = "exited" ]; then
+              echo "FAIL: backend exited during cold-start (restart loop scenario)"
+              docker logs picpeak-smoke-bk
+              echo "--- error.log ---"
+              cat smoke-mounts/logs/error.log 2>/dev/null || echo "(no error.log)"
+              exit 1
+            fi
+            if [ "$health" = "healthy" ]; then
+              echo "backend healthy after ${i}s"
+              exit 0
+            fi
+            sleep 1
+          done
+          echo "FAIL: backend did not become healthy in 120s"
+          docker ps -a
+          docker logs picpeak-smoke-bk
+          exit 1
+
+      - name: Verify chown happened (container view)
+        run: |
+          # All three dirs should now be owned by nodejs (UID 1001).
+          # If the entrypoint's self-chown branch didn't fire, they'd
+          # still be owned by the runner UID and node would have hit
+          # EACCES creating storage subdirs.
+          for d in /app/storage /app/data /app/logs; do
+            owner_uid=$(docker exec picpeak-smoke-bk stat -c '%u' "$d")
+            if [ "$owner_uid" != "1001" ]; then
+              echo "FAIL: $d is owned by UID $owner_uid (expected 1001)"
+              exit 1
+            fi
+            echo "ok: $d owned by UID $owner_uid"
+          done
+
+      - name: Verify app is actually serving
+        run: |
+          # /health is what docker's HEALTHCHECK polls, but hit it
+          # directly to confirm the response shape matches what the
+          # frontend + reverse proxy expect.
+          body=$(docker exec picpeak-smoke-bk wget -qO- http://localhost:3000/health)
+          echo "/health => $body"
+          echo "$body" | grep -q '"status":"ok"' || {
+            echo "FAIL: /health did not return status:ok"
+            exit 1
+          }
+
+      - name: Verify node runs as nodejs (not root)
+        run: |
+          # dumb-init runs as root (PID 1), node must be running as
+          # nodejs (UID 1001) — if su-exec drop didn't happen the app
+          # would be running as root which is the security regression
+          # we're guarding against. Alpine ships BusyBox ps, which
+          # doesn't support `-p PID` or pgrep, so list + awk instead.
+          user=$(docker exec picpeak-smoke-bk ps -o user,comm | awk '$2=="node" {print $1; exit}')
+          if [ "$user" != "nodejs" ]; then
+            echo "FAIL: node running as '$user' (expected nodejs)"
+            docker exec picpeak-smoke-bk ps -o pid,user,comm
+            exit 1
+          fi
+          echo "ok: node running as $user"
+
+      - name: Verify no restart loop
+        run: |
+          restart_count=$(docker inspect -f '{{.RestartCount}}' picpeak-smoke-bk)
+          if [ "$restart_count" -gt 0 ]; then
+            echo "FAIL: container restarted $restart_count time(s) — install loop bug returning"
+            docker logs picpeak-smoke-bk
+            exit 1
+          fi
+          echo "ok: 0 restarts"
+
+      # Restart with `--user 5005:5005` (no root, can't chown) against
+      # bind mounts owned by 1000 — entrypoint must fail loud with the
+      # actionable preflight error, not silently restart-loop.
+      - name: Verify preflight fails loud on unwritable mounts
+        run: |
+          docker rm -f picpeak-smoke-bk2 2>/dev/null || true
+          set +e
+          out=$(docker run --rm --user 5005:5005 --network picpeak-smoke \
+            -e NODE_ENV=production -e JWT_SECRET=x \
+            -e DB_HOST=picpeak-smoke-pg -e DB_USER=picpeak \
+            -e DB_PASSWORD=smokepass -e DB_NAME=picpeak_prod \
+            -e STORAGE_PATH=/app/storage \
+            -v "$PWD/smoke-mounts/storage:/app/storage" \
+            -v "$PWD/smoke-mounts/data:/app/data" \
+            -v "$PWD/smoke-mounts/logs:/app/logs" \
+            picpeak-backend:smoke 2>&1)
+          rc=$?
+          set -e
+          echo "$out"
+          if [ $rc -eq 0 ]; then
+            echo "FAIL: preflight should have exited non-zero"
+            exit 1
+          fi
+          echo "$out" | grep -q "is not writable by UID 5005" || {
+            echo "FAIL: preflight error message missing or wrong"
+            exit 1
+          }
+          echo "ok: preflight failed loud with actionable error"
+
+      - name: Cleanup
+        if: always()
+        run: |
+          docker rm -f picpeak-smoke-bk picpeak-smoke-bk2 picpeak-smoke-pg 2>/dev/null || true
+          docker network rm picpeak-smoke 2>/dev/null || true

.github/workflows/pr-title-lint.yml

@@ -0,0 +1,36 @@
+name: PR Title Lint
+
+# Release Please derives version bumps and the changelog from Conventional
+# Commit prefixes (feat:, fix:, ...). PRs whose title/commits use other
+# conventions (e.g. gitmoji) are silently ignored, so their changes ship
+# without a version bump or a changelog entry. This check fails a PR whose
+# title is not a valid Conventional Commit so the release stays automated.
+
+on:
+  pull_request_target:
+    types: [opened, edited, synchronize, reopened]
+
+permissions:
+  pull-requests: read
+
+jobs:
+  lint-pr-title:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Validate PR title is a Conventional Commit
+        uses: amannn/action-semantic-pull-request@v5
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        with:
+          types: |
+            feat
+            fix
+            perf
+            revert
+            docs
+            style
+            chore
+            refactor
+            test
+            build
+            ci

.github/workflows/release-please-beta.yml

@@ -0,0 +1,88 @@
+name: Release Please (Beta)
+
+on:
+  push:
+    branches: [main]
+
+permissions:
+  contents: write
+  pull-requests: write
+
+jobs:
+  release-please:
+    runs-on: ubuntu-latest
+    outputs:
+      release_created: ${{ steps.release.outputs.release_created }}
+      tag_name: ${{ steps.release.outputs.tag_name }}
+      version: ${{ steps.release.outputs.version }}
+    steps:
+      - name: Run Release Please
+        uses: googleapis/release-please-action@v4
+        id: release
+        with:
+          # A dedicated token (fine-grained PAT) makes the release PR run CI
+          # automatically (no "workflows awaiting approval") and lets it be
+          # merged without a manual review. Falls back to GITHUB_TOKEN so the
+          # workflow still works before the secret is added (#719).
+          token: ${{ secrets.RELEASE_PLEASE_TOKEN || secrets.GITHUB_TOKEN }}
+          config-file: release-please-config-beta.json
+          manifest-file: .release-please-manifest-beta.json
+          target-branch: main
+
+      # Auto-approve + enable auto-merge on the open release PR so betas publish
+      # with no manual clicks. Approval uses GITHUB_TOKEN (github-actions[bot]) —
+      # a different identity than the PR author (RELEASE_PLEASE_TOKEN) — so it is
+      # a valid review (requires the org's "Allow GitHub Actions to approve pull
+      # requests" + the repo's "Allow auto-merge"). Only meaningful when a PAT is
+      # set: without it the PR is bot-authored and can't be self-approved, so we
+      # skip and leave today's manual flow. Best-effort — never blocks the run.
+      - name: Auto-approve and enable auto-merge on the release PR
+        if: ${{ steps.release.outputs.release_created != 'true' }}
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          RELEASE_PAT: ${{ secrets.RELEASE_PLEASE_TOKEN }}
+          # This job has no checkout, so gh can't infer the repo from a git
+          # remote — set it explicitly (same pattern as whatsnew, 2a5f0a8).
+          GH_REPO: ${{ github.repository }}
+        run: |
+          if [ -z "$RELEASE_PAT" ]; then
+            echo "RELEASE_PLEASE_TOKEN not set — skipping auto-merge (manual review still required)."
+            exit 0
+          fi
+          pr=$(gh pr list --head release-please--branches--main --state open --json number --jq '.[0].number // empty')
+          if [ -n "$pr" ]; then
+            # Approve as github-actions[bot] (GITHUB_TOKEN) — a different identity
+            # than the PR author (the PAT) — so it counts as a valid review.
+            gh pr review "$pr" --approve --body "Automated approval — release-please version bump + changelog (#719)." || true
+            # Enable auto-merge as the PAT so the eventual merge commit is
+            # attributed to a real identity. If enabled via GITHUB_TOKEN the merge
+            # push is suppressed by recursion prevention and the follow-up run that
+            # cuts the tag/release never fires (#719).
+            GH_TOKEN="$RELEASE_PAT" gh pr merge "$pr" --squash --auto || true
+          else
+            echo "No open release PR to auto-merge."
+          fi
+
+      - name: Output Release Info
+        if: ${{ steps.release.outputs.release_created }}
+        run: |
+          echo "## Beta Release Created!" >> $GITHUB_STEP_SUMMARY
+          echo "" >> $GITHUB_STEP_SUMMARY
+          echo "**Tag:** ${{ steps.release.outputs.tag_name }}" >> $GITHUB_STEP_SUMMARY
+          echo "**Version:** ${{ steps.release.outputs.version }}" >> $GITHUB_STEP_SUMMARY
+          echo "" >> $GITHUB_STEP_SUMMARY
+          echo "Docker images will be built and tagged with this beta version." >> $GITHUB_STEP_SUMMARY
+
+  # Best-effort "What's New" highlights on the freshly-created release. Runs in
+  # this same workflow run (not a `release:` trigger) because release-please
+  # creates the release with GITHUB_TOKEN, which never starts new workflow runs.
+  whatsnew:
+    needs: release-please
+    if: ${{ needs.release-please.outputs.release_created }}
+    permissions:
+      contents: write   # edit the release body
+      models: read       # GitHub Models (free tier)
+    uses: ./.github/workflows/whatsnew-highlights.yml
+    with:
+      tag: ${{ needs.release-please.outputs.tag_name }}
+

.github/workflows/release-please.yml

@@ -0,0 +1,59 @@
+name: Release Please
+
+on:
+  push:
+    branches: [stable]
+
+permissions:
+  contents: write
+  pull-requests: write
+
+jobs:
+  release-please:
+    runs-on: ubuntu-latest
+    outputs:
+      release_created: ${{ steps.release.outputs.release_created }}
+      tag_name: ${{ steps.release.outputs.tag_name }}
+      version: ${{ steps.release.outputs.major }}.${{ steps.release.outputs.minor }}.${{ steps.release.outputs.patch }}
+    steps:
+      - name: Run Release Please
+        uses: googleapis/release-please-action@v4
+        id: release
+        with:
+          # Dedicated token so the release PR runs CI + can auto-merge without a
+          # manual review. Falls back to GITHUB_TOKEN before the secret is set (#719).
+          token: ${{ secrets.RELEASE_PLEASE_TOKEN || secrets.GITHUB_TOKEN }}
+          config-file: release-please-config.json
+          manifest-file: .release-please-manifest.json
+          target-branch: stable
+
+      # NOTE: stable release PRs are intentionally NOT auto-merged here
+      # anymore. Fixes accumulate in the rolling release PR and are cut as
+      # ONE patch version per day by release-stable-daily.yml (18:00 UTC,
+      # or on demand via workflow_dispatch / a manual merge of the release
+      # PR). Beta keeps instant releases — see release-please-beta.yml —
+      # because same-day reporter verification depends on it.
+
+      - name: Output Release Info
+        if: ${{ steps.release.outputs.release_created }}
+        run: |
+          echo "## Release Created! " >> $GITHUB_STEP_SUMMARY
+          echo "" >> $GITHUB_STEP_SUMMARY
+          echo "**Tag:** ${{ steps.release.outputs.tag_name }}" >> $GITHUB_STEP_SUMMARY
+          echo "**Version:** ${{ steps.release.outputs.major }}.${{ steps.release.outputs.minor }}.${{ steps.release.outputs.patch }}" >> $GITHUB_STEP_SUMMARY
+          echo "" >> $GITHUB_STEP_SUMMARY
+          echo "Docker images will be built and tagged with this version." >> $GITHUB_STEP_SUMMARY
+
+  # Best-effort "What's New" highlights on the freshly-created release. Runs in
+  # this same workflow run (not a `release:` trigger) because release-please
+  # creates the release with GITHUB_TOKEN, which never starts new workflow runs.
+  whatsnew:
+    needs: release-please
+    if: ${{ needs.release-please.outputs.release_created }}
+    permissions:
+      contents: write   # edit the release body
+      models: read       # GitHub Models (free tier)
+    uses: ./.github/workflows/whatsnew-highlights.yml
+    with:
+      tag: ${{ needs.release-please.outputs.tag_name }}
+

.github/workflows/release-stable-daily.yml

@@ -0,0 +1,86 @@
+name: Cut Stable Release (daily batch)
+
+# Stable fixes accumulate in release-please's rolling release PR instead of
+# each cutting its own patch version (the old per-merge auto-merge produced
+# e.g. 3.45.8 AND 3.45.9 on the same day). This workflow merges the open
+# stable release PR once a day, so a day of N bugfixes ships as ONE version
+# with all N changelog entries — and one Docker build instead of N.
+#
+# - schedule only fires from the default branch (main); the stable copy of
+#   this file is inert and exists to keep the branches in sync.
+# - Need a release NOW? Run this via workflow_dispatch, or merge the
+#   release PR by hand — the schedule is a default, not a gate.
+# - Approval/merge mechanics mirror the old inline step (#719): approve as
+#   github-actions[bot] (GITHUB_TOKEN, a valid distinct reviewer), enable
+#   auto-merge as the PAT so the merge attributes to a real identity and
+#   triggers the tag-cutting run. --auto waits for green checks.
+
+on:
+  schedule:
+    - cron: '0 18 * * *'
+  workflow_dispatch:
+
+permissions:
+  contents: write
+  pull-requests: write
+
+jobs:
+  merge-stable-release-pr:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Approve and enable auto-merge on the open stable release PR
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          RELEASE_PAT: ${{ secrets.RELEASE_PLEASE_TOKEN }}
+          # No checkout — set the repo explicitly so gh works without a
+          # git remote (same pattern as whatsnew, 2a5f0a8).
+          GH_REPO: ${{ github.repository }}
+        run: |
+          if [ -z "$RELEASE_PAT" ]; then
+            echo "RELEASE_PLEASE_TOKEN not set — skipping (manual review required)."
+            exit 0
+          fi
+          # Strict selection (review P1): this job runs daily even without a
+          # stable push, and `gh pr list --head` matches the branch NAME only
+          # — a fork PR can spoof `release-please--branches--stable`. Pin the
+          # base to stable AND require a same-repo head (isCrossRepository
+          # == false); a fork PR is cross-repository, so it can never be
+          # picked and auto-merged with the privileged PAT.
+          pr=$(gh pr list \
+            --base stable \
+            --head release-please--branches--stable \
+            --state open \
+            --json number,isCrossRepository \
+            --jq '[.[] | select(.isCrossRepository == false)] | .[0].number // empty')
+          if [ -z "$pr" ]; then
+            echo "No open same-repo stable release PR — nothing to cut today."
+            exit 0
+          fi
+          # Approve is tolerant — a pre-existing approval already satisfies
+          # branch protection and re-approving can return non-zero.
+          gh pr review "$pr" --approve --body "Automated approval — daily stable release batch (release-please version bump + changelog)." || echo "::warning::approve returned non-zero (PR may already be approved)"
+          # But the auto-merge enable is the load-bearing step: this scheduled
+          # job is the ONLY automatic stable cut, so DON'T swallow its failure
+          # (review P2) — an expired/under-scoped PAT would otherwise stop
+          # releases while the workflow stays green.
+          GH_TOKEN="$RELEASE_PAT" gh pr merge "$pr" --squash --auto
+          # `gh pr merge --auto` merges IMMEDIATELY when the required checks
+          # are already green — the normal case at 18:00, since the fixes
+          # merged hours earlier and CI passed. So success is EITHER the PR is
+          # already merged OR an auto-merge request is now pending; only a PR
+          # that is still open with no auto-merge request is a real failure
+          # (expired/under-scoped PAT) worth failing the job on (review round 2).
+          # One snapshot of both fields (review round 3): querying state and
+          # autoMergeRequest separately races — auto-merge can complete
+          # between the two calls, so the first sees OPEN and the second sees
+          # the request already cleared on the now-merged PR → false failure.
+          read -r state automerge < <(gh pr view "$pr" --json state,autoMergeRequest \
+            --jq '[.state, (.autoMergeRequest != null)] | @tsv')
+          if [ "$state" = "MERGED" ]; then
+            echo "Stable release PR #$pr merged immediately (checks were already green)."
+          elif [ "$automerge" = "true" ]; then
+            echo "Auto-merge enabled on stable release PR #$pr — merges when checks are green."
+          else
+            echo "::error::stable release PR #$pr is still open with no auto-merge — check RELEASE_PLEASE_TOKEN scope/expiry."
+            exit 1
+          fi

.github/workflows/release.yml

@@ -1,108 +0,0 @@
-name: Create Release
-
-on:
-  push:
-    branches:
-      - main
-    paths:
-      - 'frontend/package.json'
-      - 'backend/package.json'
-
-jobs:
-  check-version-change:
-    runs-on: ubuntu-latest
-    outputs:
-      version_changed: ${{ steps.check.outputs.changed }}
-      new_version: ${{ steps.check.outputs.version }}
-    
-    steps:
-      - name: Checkout code
-        uses: actions/checkout@v4
-        with:
-          fetch-depth: 2
-
-      - name: Check if version changed
-        id: check
-        run: |
-          # Get current versions
-          FRONTEND_VERSION=$(node -p "require('./frontend/package.json').version")
-          BACKEND_VERSION=$(node -p "require('./backend/package.json').version")
-          
-          # Get previous versions
-          git checkout HEAD~1
-          PREV_FRONTEND_VERSION=$(node -p "require('./frontend/package.json').version" 2>/dev/null || echo "0.0.0")
-          PREV_BACKEND_VERSION=$(node -p "require('./backend/package.json').version" 2>/dev/null || echo "0.0.0")
-          
-          # Check if versions changed
-          if [[ "$FRONTEND_VERSION" != "$PREV_FRONTEND_VERSION" ]] || [[ "$BACKEND_VERSION" != "$PREV_BACKEND_VERSION" ]]; then
-            echo "changed=true" >> $GITHUB_OUTPUT
-            echo "version=$FRONTEND_VERSION" >> $GITHUB_OUTPUT
-          else
-            echo "changed=false" >> $GITHUB_OUTPUT
-          fi
-
-  create-release:
-    needs: check-version-change
-    if: needs.check-version-change.outputs.version_changed == 'true'
-    runs-on: ubuntu-latest
-    permissions:
-      contents: write
-      
-    steps:
-      - name: Checkout code
-        uses: actions/checkout@v4
-
-      - name: Generate Changelog
-        id: changelog
-        run: |
-          # Get commits since last tag
-          LAST_TAG=$(git describe --tags --abbrev=0 2>/dev/null || echo "")
-          if [[ -z "$LAST_TAG" ]]; then
-            COMMITS=$(git log --oneline)
-          else
-            COMMITS=$(git log ${LAST_TAG}..HEAD --oneline)
-          fi
-          
-          # Format changelog
-          echo "## What's Changed" > changelog.md
-          echo "" >> changelog.md
-          
-          # Group commits by type
-          echo "### Features" >> changelog.md
-          echo "$COMMITS" | grep -E "^[a-f0-9]+ feat:" | sed 's/^[a-f0-9]+ /- /' >> changelog.md || echo "*No new features*" >> changelog.md
-          
-          echo "" >> changelog.md
-          echo "### Bug Fixes" >> changelog.md
-          echo "$COMMITS" | grep -E "^[a-f0-9]+ fix:" | sed 's/^[a-f0-9]+ /- /' >> changelog.md || echo "*No bug fixes*" >> changelog.md
-          
-          echo "" >> changelog.md
-          echo "### Other Changes" >> changelog.md
-          echo "$COMMITS" | grep -vE "^[a-f0-9]+ (feat|fix):" | sed 's/^[a-f0-9]+ /- /' >> changelog.md || echo "*No other changes*" >> changelog.md
-          
-          # Save changelog
-          echo "changelog<<EOF" >> $GITHUB_OUTPUT
-          cat changelog.md >> $GITHUB_OUTPUT
-          echo "EOF" >> $GITHUB_OUTPUT
-
-      - name: Create Release
-        uses: softprops/action-gh-release@v1
-        with:
-          tag_name: v${{ needs.check-version-change.outputs.new_version }}
-          name: Release v${{ needs.check-version-change.outputs.new_version }}
-          body: |
-            ## PicPeak v${{ needs.check-version-change.outputs.new_version }}
-            
-            ${{ steps.changelog.outputs.changelog }}
-            
-            ### Docker Images
-            
-            To use this release with Docker:
-            ```bash
-            docker pull ghcr.io/${{ github.repository }}/frontend:v${{ needs.check-version-change.outputs.new_version }}
-            docker pull ghcr.io/${{ github.repository }}/backend:v${{ needs.check-version-change.outputs.new_version }}
-            ```
-            
-            Or use the `latest` tag for the most recent version.
-          draft: false
-          prerelease: false
-          generate_release_notes: true

.github/workflows/schema-drift.yml

@@ -0,0 +1,186 @@
+name: Schema drift (#530)
+
+# Verifies that `migrate:safe` can recover a DB that's been seeded only
+# by `initializeDatabase()` — the recovery scenario where the migrations
+# tracking table is empty but the schema already has the modern bootstrap.
+#
+# This is NOT how production reaches its state on normal installs or
+# upgrades. The scenario only fires when:
+#   - A backup was restored that captured tables but not the migrations
+#     table (manifest divergence),
+#   - Someone manually invoked initializeDatabase() outside the migration
+#     runner (recovery / debugging),
+#   - The DB was moved between systems and the migrations table was not
+#     copied along.
+#
+# When `detectExistingSchema()` sees the modern-bootstrap fingerprint
+# (photo_categories + cms_pages tables) but an empty migrations table,
+# it treats it as an "existing deployment" — which runs the legacy
+# chain first. Legacy/008 renames email_templates.subject → subject_en,
+# but core/029 (which runs later in this chain) inserts email templates
+# referencing the pre-rename column name. The chain dies with a
+# "column subject does not exist" error.
+#
+# Fix (in the same PR as this workflow): when the modern-bootstrap
+# fingerprint is detected, mark all legacy migrations as applied so the
+# chain matches what a fresh install runs — only core/*, in order.
+#
+# This workflow boots the failing scenario from scratch on every PR
+# that touches the migrations or db.js, so any future migration with
+# the same shape is caught before merge.
+
+on:
+  # No `paths:` filter — branch protection on `main` + `stable` lists
+  # `upgrade-from-bootstrap` as a REQUIRED check. A path-filtered
+  # trigger that skipped on unrelated PRs would leave the required
+  # check "missing" forever, blocking every PR that doesn't touch
+  # migrations. The ~75-second cost on every PR buys an unconditional
+  # safety net. (Branches also updated post-#669 rename: beta → main,
+  # old main → stable.)
+  push:
+    branches: [main, stable]
+  pull_request:
+    branches: [main, stable]
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+jobs:
+  upgrade-from-bootstrap:
+    runs-on: ubuntu-latest
+    timeout-minutes: 10
+
+    services:
+      postgres:
+        image: postgres:15-alpine
+        env:
+          POSTGRES_USER: picpeak
+          POSTGRES_PASSWORD: testpass
+          POSTGRES_DB: picpeak_drift
+        options: >-
+          --health-cmd "pg_isready -U picpeak -d picpeak_drift"
+          --health-interval 2s
+          --health-timeout 2s
+          --health-retries 30
+        ports:
+          - 5432:5432
+
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v4
+
+      - name: Set up Node.js
+        uses: actions/setup-node@v4
+        with:
+          node-version: '22'
+          cache: 'npm'
+          cache-dependency-path: backend/package-lock.json
+
+      - name: Install backend deps
+        working-directory: ./backend
+        run: npm ci
+
+      # Step 1: simulate the recovery state — DB has the modern bootstrap
+      # (post-initializeDatabase) but no migrations recorded. Calling
+      # initializeDatabase() directly outside the migration runner is the
+      # one-line repro for backup-restore-lost-migrations and manual-
+      # invocation paths.
+      - name: Seed DB with initializeDatabase() only
+        working-directory: ./backend
+        env:
+          NODE_ENV: production
+          DATABASE_CLIENT: pg
+          DB_HOST: localhost
+          DB_PORT: 5432
+          DB_USER: picpeak
+          DB_PASSWORD: testpass
+          DB_NAME: picpeak_drift
+        run: |
+          node -e "require('./src/database/db').initializeDatabase().then(() => { console.log('bootstrap ok'); process.exit(0); }).catch(e => { console.error('bootstrap FAILED:', e.message); process.exit(1); })"
+
+      # Sanity-check the recovery shape before migrate:safe runs. If
+      # initializeDatabase() ever stops producing photo_categories +
+      # cms_pages, the fingerprint check would silently no-op and this
+      # workflow would lose its teeth — assert the precondition.
+      - name: Assert recovery-state fingerprint
+        env:
+          PGPASSWORD: testpass
+        run: |
+          installed=$(psql -h localhost -U picpeak -d picpeak_drift -tAc "SELECT count(*) FROM pg_tables WHERE schemaname='public' AND tablename IN ('photo_categories', 'cms_pages')")
+          if [ "$installed" != "2" ]; then
+            echo "FAIL: expected photo_categories + cms_pages from initializeDatabase(); got $installed."
+            psql -h localhost -U picpeak -d picpeak_drift -c "SELECT tablename FROM pg_tables WHERE schemaname='public' ORDER BY tablename"
+            exit 1
+          fi
+          # initializeDatabase() doesn't create the `migrations` tracking
+          # table — that's the migrate:safe runner's job. So in the recovery
+          # scenario, the table either (a) doesn't exist yet or (b) exists
+          # but is empty (e.g. someone created it but didn't populate it).
+          # Both are valid recovery states; check via to_regclass first so
+          # we don't parse a SELECT against a nonexistent table.
+          has_migrations_table=$(psql -h localhost -U picpeak -d picpeak_drift -tAc "SELECT to_regclass('public.migrations')::text")
+          if [ -z "$has_migrations_table" ]; then
+            migrations_count=0
+          else
+            migrations_count=$(psql -h localhost -U picpeak -d picpeak_drift -tAc "SELECT count(*) FROM migrations")
+          fi
+          if [ "$migrations_count" != "0" ]; then
+            echo "FAIL: migrations table should be empty for the recovery scenario; has $migrations_count rows."
+            exit 1
+          fi
+          echo "ok: recovery state confirmed (bootstrap tables present, migrations table empty or absent)."
+
+      # Step 2: run migrate:safe — the test. Before #530's fix in
+      # detectExistingSchema, this died at core/029 with a "column
+      # subject does not exist" error. After the fix, it should complete
+      # cleanly with every migration either applied or marked.
+      - name: Run migrate:safe against the recovery state
+        working-directory: ./backend
+        env:
+          NODE_ENV: production
+          DATABASE_CLIENT: pg
+          DB_HOST: localhost
+          DB_PORT: 5432
+          DB_USER: picpeak
+          DB_PASSWORD: testpass
+          DB_NAME: picpeak_drift
+        run: npm run migrate:safe
+
+      # Step 3: schema-shape assertion. A fresh install through migrate:
+      # safe produces 48 tables; the recovery scenario should converge
+      # to the same number. Off-by-one is fine but a 10+ table delta
+      # means a migration silently bailed in the recovery path.
+      - name: Assert final schema matches fresh-install shape
+        env:
+          PGPASSWORD: testpass
+        run: |
+          tables=$(psql -h localhost -U picpeak -d picpeak_drift -tAc "SELECT count(*) FROM pg_tables WHERE schemaname='public'")
+          echo "Final table count: $tables"
+          # Allow a small drift window — exact count creeps over time as
+          # new migrations land; tight pin would force a workflow edit
+          # on every schema PR. 40+ is a healthy floor that catches the
+          # original bug (which left 17 tables) while staying robust to
+          # forward changes.
+          if [ "$tables" -lt 40 ]; then
+            echo "FAIL: too few tables ($tables) — migrate:safe likely bailed mid-chain."
+            psql -h localhost -U picpeak -d picpeak_drift -c "SELECT tablename FROM pg_tables WHERE schemaname='public' ORDER BY tablename"
+            exit 1
+          fi
+          echo "ok: schema converged to a fresh-install-equivalent shape."
+
+      # Step 4: verify the legacy migrations were all marked applied
+      # (rather than silently bailing inside the chain). The fix in
+      # detectExistingSchema marks legacy/* when the modern bootstrap
+      # is detected — confirm the markings actually landed.
+      - name: Assert legacy migrations marked applied
+        env:
+          PGPASSWORD: testpass
+        run: |
+          legacy_count=$(psql -h localhost -U picpeak -d picpeak_drift -tAc "SELECT count(*) FROM migrations WHERE filename LIKE '008_%' OR filename LIKE '009_%' OR filename LIKE '013_%' OR filename LIKE '019_%' OR filename LIKE '020_%' OR filename LIKE '026_%' OR filename LIKE '028_%'")
+          if [ "$legacy_count" -lt 7 ]; then
+            echo "FAIL: legacy migrations not marked applied ($legacy_count of 7 expected)."
+            psql -h localhost -U picpeak -d picpeak_drift -c "SELECT filename FROM migrations WHERE filename LIKE '0%' ORDER BY filename"
+            exit 1
+          fi
+          echo "ok: legacy migrations marked applied by detectExistingSchema."

.github/workflows/tests.yml

@@ -0,0 +1,115 @@
+name: Tests
+
+# Runs the backend Jest suite and the frontend Vitest suite on every PR.
+# Both suites already exist and cover the CRM service layer (quoteService,
+# contractService, invoiceService.*, customerHoursService, eventService.
+# calendar) plus the photo / settings / OG / auth surface — wiring them
+# into CI makes regressions visible at PR time instead of post-merge.
+#
+# Six backend suites are excluded via --testPathIgnorePatterns. They
+# fail on `upstream/beta` too (pre-existing mock/infra issues, NOT CRM
+# regressions). Excluding them here keeps CI green from day 1; revisit
+# each individually as its own fix.
+#
+# Triggers on any change that could affect either suite. The backend
+# job intentionally omits frontend paths and vice versa so unrelated
+# PRs don't pay both build costs.
+
+on:
+  push:
+    branches: [main, beta, stable]
+  pull_request:
+    branches: [main, beta, stable]
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+jobs:
+  backend:
+    runs-on: ubuntu-latest
+    timeout-minutes: 10
+
+    # The .picpeak restore suites gate their real-Postgres cases behind
+    # PICPEAK_PG_TEST_URL and `describe.skip` themselves out when it is
+    # unset — so until now they never ran here. That hid the half that
+    # matters: sequence resync, operator/role preservation across a
+    # cross-instance restore, and (with #1041) whether a SQLite-shaped
+    # row actually lands in Postgres with the right STORED VALUES rather
+    # than merely not throwing. Everything else in the suite still runs
+    # on SQLite; this service only un-gates those cases.
+    services:
+      postgres:
+        image: postgres:15-alpine
+        env:
+          POSTGRES_USER: picpeak
+          POSTGRES_PASSWORD: testpass
+          POSTGRES_DB: picpeak_test
+        options: >-
+          --health-cmd "pg_isready -U picpeak -d picpeak_test"
+          --health-interval 2s
+          --health-timeout 2s
+          --health-retries 30
+        ports:
+          - 5432:5432
+
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v4
+
+      - name: Set up Node.js
+        uses: actions/setup-node@v4
+        with:
+          node-version: '22'
+          cache: 'npm'
+          cache-dependency-path: backend/package-lock.json
+
+      - name: Install backend deps
+        working-directory: ./backend
+        run: npm ci
+
+      - name: Run Jest suite
+        working-directory: ./backend
+        env:
+          # backupService tests would otherwise try a real S3 round-trip.
+          # The S3 path itself is covered separately by the integration
+          # suite when MinIO is provisioned.
+          SKIP_S3_TESTS: 'true'
+          # Un-gates the real-Postgres cases in the .picpeak restore suites
+          # (see the `services:` note above). Absent it they silently skip.
+          PICPEAK_PG_TEST_URL: 'postgres://picpeak:testpass@127.0.0.1:5432/picpeak_test'
+        run: |
+          # Excluded suites — fail on upstream/beta too, tracked
+          # separately as test-infra debt:
+          #   adminSettings.logo                  — supertest fixture
+          #   integration/adminPhotos.reference   — supertest fixture
+          #   integration/webhookDelivery         — supertest fixture
+          #   services/backupService.enhanced     — knex mock chain
+          #   routes/__tests__/adminAuth          — supertest fixture
+          # (adminNotifications was excluded; #597 fix re-enables it.)
+          npx jest \
+            --testPathIgnorePatterns='/node_modules/|adminSettings\.logo\.test|integration/adminPhotos\.reference|integration/webhookDelivery|backupService\.enhanced|routes/__tests__/adminAuth' \
+            --ci
+
+  frontend:
+    runs-on: ubuntu-latest
+    timeout-minutes: 10
+
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v4
+
+      - name: Set up Node.js
+        uses: actions/setup-node@v4
+        with:
+          node-version: '22'
+          cache: 'npm'
+          cache-dependency-path: frontend/package-lock.json
+
+      - name: Install frontend deps
+        working-directory: ./frontend
+        run: npm ci
+
+      - name: Run Vitest suite
+        working-directory: ./frontend
+        run: npm test -- --run

.github/workflows/version-bump.yml

@@ -1,107 +0,0 @@
-name: Automatic Version Bump
-
-on:
-  push:
-    branches:
-      - main
-  workflow_dispatch:
-    inputs:
-      version_type:
-        description: 'Version bump type'
-        required: true
-        default: 'patch'
-        type: choice
-        options:
-          - patch
-          - minor
-          - major
-
-jobs:
-  version-bump:
-    runs-on: ubuntu-latest
-    permissions:
-      contents: write
-      pull-requests: write
-    
-    steps:
-      - name: Checkout code
-        uses: actions/checkout@v4
-        with:
-          fetch-depth: 0
-          token: ${{ secrets.GITHUB_TOKEN }}
-
-      - name: Setup Node.js
-        uses: actions/setup-node@v4
-        with:
-          node-version: '20'
-          
-      - name: Configure Git
-        run: |
-          git config --global user.name "GitHub Actions Bot"
-          git config --global user.email "actions@github.com"
-
-      - name: Determine version type
-        id: version_type
-        run: |
-          if [[ "${{ github.event_name }}" == "workflow_dispatch" ]]; then
-            echo "type=${{ github.event.inputs.version_type }}" >> $GITHUB_OUTPUT
-          else
-            # Auto-detect version type based on commit message
-            COMMIT_MSG="${{ github.event.head_commit.message }}"
-            if [[ "$COMMIT_MSG" == *"BREAKING CHANGE"* ]] || [[ "$COMMIT_MSG" == *"!"* ]]; then
-              echo "type=major" >> $GITHUB_OUTPUT
-            elif [[ "$COMMIT_MSG" == *"feat:"* ]] || [[ "$COMMIT_MSG" == *"feat("* ]]; then
-              echo "type=minor" >> $GITHUB_OUTPUT
-            else
-              echo "type=patch" >> $GITHUB_OUTPUT
-            fi
-          fi
-
-      - name: Bump Frontend Version
-        id: frontend_version
-        working-directory: ./frontend
-        run: |
-          npm version ${{ steps.version_type.outputs.type }} --no-git-tag-version
-          NEW_VERSION=$(node -p "require('./package.json').version")
-          echo "version=$NEW_VERSION" >> $GITHUB_OUTPUT
-          
-      - name: Bump Backend Version
-        id: backend_version
-        working-directory: ./backend
-        run: |
-          npm version ${{ steps.version_type.outputs.type }} --no-git-tag-version
-          NEW_VERSION=$(node -p "require('./package.json').version")
-          echo "version=$NEW_VERSION" >> $GITHUB_OUTPUT
-
-      - name: Update Frontend VersionInfo component
-        run: |
-          VERSION=${{ steps.frontend_version.outputs.version }}
-          sed -i "s/const FRONTEND_VERSION = '[^']*'/const FRONTEND_VERSION = '$VERSION'/" frontend/src/components/admin/VersionInfo.tsx
-
-      - name: Create Pull Request
-        uses: peter-evans/create-pull-request@v5
-        with:
-          token: ${{ secrets.GITHUB_TOKEN }}
-          commit-message: "chore: bump version to ${{ steps.frontend_version.outputs.version }}"
-          title: "chore: bump version to ${{ steps.frontend_version.outputs.version }}"
-          body: |
-            ## Version Bump
-            
-            This PR automatically bumps the version numbers:
-            - Frontend: `${{ steps.frontend_version.outputs.version }}`
-            - Backend: `${{ steps.backend_version.outputs.version }}`
-            
-            ### Version Type: ${{ steps.version_type.outputs.type }}
-            
-            ### Files Changed:
-            - `frontend/package.json`
-            - `backend/package.json`
-            - `frontend/src/components/admin/VersionInfo.tsx`
-            
-            ---
-            *This PR was automatically created by the version bump workflow.*
-          branch: version-bump-${{ steps.frontend_version.outputs.version }}
-          delete-branch: true
-          labels: |
-            version-bump
-            automated

.github/workflows/whatsnew-highlights.yml

@@ -0,0 +1,98 @@
+# What's New highlights — GitHub Models release step (reusable)
+#
+# Called by the release-please workflows AFTER a release is created
+# (release-please.yml for `stable`, release-please-beta.yml for `main`). It runs
+# as a job in the SAME workflow run rather than on its own `release: published`
+# trigger, because release-please creates the release with the default
+# GITHUB_TOKEN and GitHub does not start new workflow runs from token-generated
+# events — a standalone `release:` workflow would simply never fire.
+#
+# What it does: condenses the new release's "### Features" into <=8 short
+# bullets via GitHub Models (free tier, `models: read`) and injects a
+# `<!-- whatsnew -->` block at the top of the release notes. The app reads that
+# block (backend utils/whatsNew.parseWhatsNew) and falls back to the raw
+# Features list for releases without it — so this is purely a quality upgrade,
+# never a hard dependency. Failure is isolated by `continue-on-error` + the
+# deterministic fallback below, so it can never break a release.
+#
+# GitHub Models is OPTIONAL. If it is disabled/unavailable for the org the AI
+# step fails soft (continue-on-error) and the deterministic fallback produces
+# the bullets instead — the feature works either way, Models just polishes them.
+#
+# Validated end-to-end on a fork (extract -> openai/gpt-4o-mini -> inject into
+# real release notes; app parseWhatsNew() reads the block back).
+
+name: What's New highlights
+
+on:
+  workflow_call:
+    inputs:
+      tag:
+        description: Release tag to annotate (e.g. v2.3.0)
+        required: true
+        type: string
+
+jobs:
+  highlights:
+    runs-on: ubuntu-latest
+    permissions:
+      contents: write   # to edit the release body
+      models: read      # GitHub Models (free tier)
+    # GH_REPO at job scope so every `gh` call targets the right repo without
+    # needing an actions/checkout step. Without this, `gh` falls back to
+    # parsing `.git/config` in the runner's empty workspace and dies with
+    # "fatal: not a git repository" — which hard-fails the whole job before
+    # any continue-on-error can save it.
+    env:
+      GH_REPO: ${{ github.repository }}
+    steps:
+      - name: Extract Features from the published release
+        id: feat
+        # Belt-and-braces: the job-level comment says "never let highlights
+        # break a release", but the original wiring only marked the AI +
+        # inject steps as continue-on-error. A hiccup here (rate limit,
+        # transient API error) would still hard-fail the job. Match the
+        # design intent and fail soft.
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          TAG: ${{ inputs.tag }}
+        run: |
+          BODY=$(gh release view "$TAG" --json body -q .body)
+          FEATURES=$(printf '%s\n' "$BODY" | awk '/^#{2,4} +Features/{f=1;next} /^#{1,4} +\S/{f=0} f')
+          { echo "features<<EOF"; printf '%s\n' "$FEATURES"; echo EOF; } >> "$GITHUB_OUTPUT"
+
+      - name: Summarize with GitHub Models
+        if: ${{ steps.feat.outputs.features != '' }}
+        id: ai
+        continue-on-error: true   # Models may be disabled/unavailable for the org; fall back deterministically below
+        uses: actions/ai-inference@v1
+        with:
+          model: openai/gpt-4o-mini   # catalog id (verified present); openai/gpt-4.1-mini or openai/gpt-5-nano also work
+          system-prompt: >
+            You write release highlights for the admins of a self-hosted
+            photo-gallery + CRM app. Given raw changelog "Features" lines, output
+            AT MOST 8 markdown bullets, each 3-4 words, user-facing, no scopes,
+            no jargon, no issue numbers. One bullet per distinct user-visible
+            feature. Output ONLY "- " bullets, nothing else.
+          prompt: ${{ steps.feat.outputs.features }}
+
+      - name: Inject the What's New block
+        if: ${{ steps.feat.outputs.features != '' }}
+        continue-on-error: true   # never let highlights break a release
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          TAG: ${{ inputs.tag }}
+          AI: ${{ steps.ai.outputs.response }}
+          FEATURES: ${{ steps.feat.outputs.features }}
+        run: |
+          BULLETS="$AI"
+          # Deterministic fallback if the model returned nothing (e.g. Models not yet enabled).
+          if [ -z "$BULLETS" ]; then
+            BULLETS=$(printf '%s\n' "$FEATURES" | head -8 \
+              | sed -E 's/^\* \*\*[^:]+:\*\* */- /; s/ \(\[[^]]*\]\([^)]*\)\)//g')
+          fi
+          BODY=$(gh release view "$TAG" --json body -q .body)
+          # Idempotent: strip any prior block before re-injecting.
+          BODY=$(printf '%s' "$BODY" | perl -0pe 's/<!--\s*whatsnew\s*-->.*?<!--\s*\/whatsnew\s*-->\n*//is')
+          gh release edit "$TAG" --notes "$(printf '<!-- whatsnew -->\n%s\n<!-- /whatsnew -->\n\n%s' "$BULLETS" "$BODY")"

.gitignore

@@ -11,6 +11,9 @@ yarn-error.log*
 .env.test.local
 .env.production.local
 
+# Docker override file
+docker-compose.override.yml
+
 # Security - Never commit credentials
 ADMIN_CREDENTIALS.txt
 ADMIN_PASSWORD_RESET.txt
@@ -48,9 +51,106 @@ coverage/
 *.tmp
 *.temp
 
+# Backup and test directories
+backups/
+test-archiver/
+
 # Keep directory structure
 !storage/events/active/.gitkeep
 !storage/events/archived/.gitkeep
 !storage/thumbnails/.gitkeep
 !data/.gitkeep
 !logs/.gitkeep
+
+# development files
+backend/.swarm/
+.claudedocs/
+backend/data/
+backend/docs/
+backend/logs/
+logs/
+# Anchored to repo root: matches the top-level runtime storage dir,
+# NOT backend/src/services/storage/ (the storage backend abstraction code).
+/storage/
+data/
+certbot/
+
+# Ignore local contributor guide copy
+AGENTS.md
+CLAUDE.md
+
+# Working/planning documents (not for release)
+BUGS_AND_FEATURES.md
+frontend/TEST_PLAN.md
+docs/REFACTORING_PLAN.md
+docs/MULTIPLE_ADMINISTRATORS_PLAN.md
+docs/*_PLAN.md
+docs/test-*.md
+docs/feature-*.md
+
+# Scaffolding documentation (local development reference)
+docs/DATABASE_SCHEMA.md
+docs/BACKEND_SERVICES.md
+docs/API_ROUTES.md
+docs/FRONTEND_ARCHITECTURE.md
+docs/DEVELOPER_ONBOARDING.md
+docs/ENVIRONMENT_VARIABLES.md
+
+# Build artifact: OpenAPI spec generated locally + synced into the
+# picpeak-docs repo. Never tracked here — the docs site at
+# docs.picpeak.app is the source of truth.
+docs/openapi.json
+docs/openapi.yaml
+
+# Local backup directory (from testing)
+backup/
+
+# Local artifacts from browser tooling
+.playwright-mcp/
+
+# Local-only E2E suite (never pushed; runs as pre-push gate on this machine)
+tests/e2e/local/
+playwright-local-results/
+e2e-test.log
+scripts/e2e-local.sh
+
+# Local SQLite files in backend
+backend/*.sqlite*
+backend/*.db
+
+# Test files and artifacts
+test-images/
+test-logo*.jpg
+test-logo*.png
+test-results/
+
+# Development docker compose
+docker-compose.dev.yml
+
+# New layout development files
+new-layouts/
+
+# Backend runtime storage (generated media, previews, thumbnails,
+# CRM/accounting documents) — never commit. Matches main: a dev instance
+# writes event photos into backend/storage/, and the narrower
+# business-docs-only rule let `git add -A` sweep them into a commit.
+backend/storage/
+
+# Python bytecode. The ML sidecar lives on main only, so this branch never
+# needed the rule — which is how a `git add -A` from a shared working tree
+# committed 16 .pyc files here in #1247.
+__pycache__/
+*.pyc
+
+# Issue / PR screenshots belong on a `screenshots/*` branch, never on main or
+# stable — that is what those branches exist for. Two landed at the repo root
+# on main in #1241 and shipped as part of the source tree; nothing stopped it.
+#
+# Anchored with a leading slash so docs/ keeps its own images.
+/issue-*.png
+/issue-*.jpg
+/screenshot-*.png
+/screenshot-*.jpg
+/*-screenshot.png
+/*-screenshot.jpg
+

.release-please-manifest-beta.json

@@ -0,0 +1,3 @@
+{
+  ".": "3.83.0-beta.0"
+}

.release-please-manifest.json

@@ -0,0 +1 @@
+{".":"3.46.12"}

CI_CD_STRATEGY.md

@@ -1,118 +0,0 @@
-# CI/CD Strategy for PicPeak
-
-## Overview
-
-This document outlines the CI/CD strategy using both Gitea Actions and Drone CI to avoid conflicts and ensure proper versioning.
-
-## Pipeline Flow
-
-### 1. Development & Testing (Gitea Actions)
-- **Trigger**: Every push to `main` or `develop` branches
-- **File**: `.gitea/workflows/test.yml`
-- **Purpose**: Run tests, linting, and basic validation
-- **Actions**:
-  - Backend linting and tests
-  - Frontend linting and build
-  - Does NOT build Docker images
-
-### 2. Version Management (Gitea Actions)
-- **Trigger**: Push to `main` branch (excluding markdown files)
-- **File**: `.gitea/workflows/version-and-release.yml`
-- **Purpose**: Automatic version incrementing
-- **Actions**:
-  1. Reads current version from `package.json`
-  2. Increments patch version (e.g., 1.0.0 → 1.0.1)
-  3. Updates both backend and frontend `package.json`
-  4. Commits the version change
-  5. Creates a git tag (e.g., `v1.0.1`)
-  6. Pushes changes and tag
-
-### 3. Docker Image Building (Drone CI)
-- **Trigger**: 
-  - Push to `main` or `develop` (builds with commit SHA)
-  - New git tags (builds release versions)
-- **File**: `.drone.yml`
-- **Purpose**: Build and push Docker images
-- **Tags Created**:
-  - `latest` - Always points to newest build
-  - `{commit-sha}` - Specific commit version
-  - `{branch}-latest` - Latest for specific branch
-  - `v1.0.1` - Specific version (on tag trigger)
-
-## Why This Strategy?
-
-1. **Separation of Concerns**:
-   - Gitea Actions handles code quality and versioning
-   - Drone CI handles Docker image building
-   - No overlap or race conditions
-
-2. **Sequential Execution**:
-   - Version bump happens first
-   - Tag creation triggers Drone
-   - Docker images are built with correct version
-
-3. **Version Consistency**:
-   - Version in `package.json` matches git tag
-   - Docker images are tagged with same version
-   - No manual version management needed
-
-## Setup Requirements
-
-1. **Gitea Actions Runner**: Must be configured and running
-2. **Drone CI**: Must be connected to your Gitea instance
-3. **Secrets**: 
-   - `GITEA_TOKEN` (optional, for pushing version commits)
-   - Docker registry credentials in Drone
-
-## Version Numbering
-
-- Format: `MAJOR.MINOR.PATCH` (e.g., 1.0.0)
-- Automatic increments: PATCH version only
-- Manual increments: Edit `package.json` for MAJOR/MINOR changes
-
-## Usage
-
-1. **Regular Development**:
-   ```bash
-   git add .
-   git commit -m "feat: add new feature"
-   git push origin main
-   ```
-   - Tests run automatically
-   - Version bumps to 1.0.1
-   - Docker images built with v1.0.1 tag
-
-2. **Major/Minor Version Change**:
-   ```bash
-   # Manually edit package.json files to 2.0.0
-   git add .
-   git commit -m "feat!: major release"
-   git push origin main
-   ```
-
-3. **Skip Version Bump**:
-   - Add `[skip ci]` to commit message
-   - Or only change markdown files
-
-## Monitoring
-
-- **Gitea Actions**: Check Actions tab in Gitea
-- **Drone CI**: Check Drone dashboard
-- **Docker Registry**: Verify images are pushed with correct tags
-
-## Troubleshooting
-
-1. **Version not incrementing**:
-   - Check Gitea Actions logs
-   - Ensure runner has push permissions
-   - Verify no `[skip ci]` in commit message
-
-2. **Docker images not building**:
-   - Check Drone CI webhook configuration
-   - Verify Drone can see the repository
-   - Check Docker registry credentials
-
-3. **Conflicts**:
-   - Never run both pipelines for same task
-   - Use branch protection to prevent direct pushes
-   - Always let automation handle versioning

CLAUDE.md

@@ -1,261 +0,0 @@
-# CLAUDE.md
-
-This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
-
-## Product Overview
-
-A secure photo sharing platform designed for weddings and events, enabling photographers to share time-limited, password-protected galleries. The platform features automatic expiration, archiving, and a scrappbook.de-inspired modern, minimalist UI.
-
-## Architecture Overview
-
-- **Backend**: Node.js/Express API with SQLite/PostgreSQL, file-based photo storage
-- **Frontend**: React SPA with scrappbook.de-style design (requires implementation)
-- **Storage**: File-based with active/archived separation
-- **Services**: Background workers for email, archiving, file watching, and expiration monitoring
-- **Analytics**: Umami integration for engagement tracking
-
-## Essential Commands
-
-### Backend Development
-```bash
-cd backend
-npm install               # Install dependencies
-npm run migrate          # Initialize database schema
-npm run dev              # Start with hot-reload (port 3001)
-npm test                 # Run Jest tests
-npm run lint             # ESLint checks
-```
-
-### Running a Single Test
-```bash
-cd backend
-npm test -- path/to/test.test.js
-npm test -- --testNamePattern="test name"
-```
-
-### Production
-```bash
-docker-compose -f docker-compose.prod.yml up -d  # Production deployment
-pm2 start ecosystem.config.js                     # Alternative: PM2 deployment
-```
-
-## Key Product Requirements (from PRD)
-
-### Core Features
-1. **File-Based System**: Drop photos in folders → automatic gallery creation
-2. **Automatic Expiration**: Default 30 days, with 7-day warning emails
-3. **Password Protection**: Secure access with customizable passwords
-4. **Automatic Archiving**: ZIP compression and storage after expiration
-5. **Email Notifications**: Creation, warning, and expiration notifications
-6. **Analytics**: Umami tracking for views, downloads, and engagement
-
-### Folder Structure
-```
-/events/
-  ├── active/
-  │   ├── wedding-smith-jones-2024-06-15/
-  │   │   ├── collages/
-  │   │   └── individual/
-  │   └── birthday-emma-2024-07-20/
-  └── archived/
-      └── wedding-smith-jones-2024-06-15.zip
-```
-
-## Frontend Implementation Requirements
-
-### Design Style (scrappbook.de-inspired)
-- **Color Palette**: Primary green (#5C8762), neutral backgrounds
-- **Typography**: Clean, modern sans-serif (Noto Sans or similar)
-- **Layout**: Minimalist, modular sections with grid-based photo displays
-- **Aesthetic**: Professional yet approachable, photographer-focused
-
-### Key Frontend Components to Build
-1. **Landing Page**: Password entry with event preview
-2. **Gallery View**: 
-   - Responsive photo grid with lazy loading
-   - Toggle between collages/individual photos
-   - Prominent expiration banner
-   - Download urgency indicators
-3. **Photo Lightbox**: Full-screen viewing with zoom
-4. **Mobile-First**: Responsive design with touch gestures
-5. **Personalization**: Dynamic theming per event type
-
-### User Experience Priorities
-- Clear expiration warnings (sticky banner)
-- One-click "Download All" for urgent galleries
-- Smooth image loading with skeleton screens
-- Intuitive navigation between photo categories
-- Professional presentation matching photographer branding
-
-## Key Architecture Patterns
-
-### Authentication Flow
-- JWT-based with separate tokens for admin and gallery access
-- Gallery tokens include event-specific claims
-- Auth middleware: `backend/src/middleware/auth.js`
-  - `adminAuth` - Admin panel protection
-  - `photoAuth` - Protected photo access  
-  - `verifyGalleryAccess` - Gallery-specific validation
-
-### Database Schema (Knex/SQLite)
-Main tables:
-- `events` - Gallery metadata with expiration, custom messages, themes
-- `photos` - Photo records linked to events
-- `access_logs` - IP-based usage tracking
-- `email_queue` - Async email processing
-- `admin_users` - Admin authentication
-
-### Service Architecture
-Background services run as separate processes:
-- **emailService**: Processes email queue with retry logic
-- **archiveService**: Creates ZIP archives of expired events
-- **expirationChecker**: Cron job for expiration warnings
-- **fileWatcher**: Monitors for new photo uploads
-
-### API Structure
-- `/api/admin/*` - Admin panel endpoints (requires adminAuth)
-- `/api/gallery/*` - Public gallery endpoints  
-- `/api/auth/*` - Authentication endpoints
-- Rate limiting: 100 req/15min (general), 5 req/15min (auth)
-
-## Critical Implementation Notes
-
-1. **Security**: All gallery access requires valid JWT with event-specific claims
-2. **Expiration**: Events auto-expire based on `expires_at`, with 7-day email warnings
-3. **Email Queue**: Async processing with retry logic, check `email_queue` table
-4. **File Processing**: Sharp library for thumbnail generation (300x300)
-5. **Frontend Status**: Only skeleton exists - requires full implementation based on PRD
-6. **Umami Analytics**: Track password entries, downloads, views, expiration warnings
-
-## Environment Variables
-
-### Backend (.env)
-- `JWT_SECRET` - Token signing
-- `ADMIN_URL`, `FRONTEND_URL` - CORS origins
-- `SMTP_*` - Email configuration
-- `DB_*` - PostgreSQL credentials (production)
-- `UMAMI_URL` - Umami instance URL (for server-side tracking)
-- `UMAMI_WEBSITE_ID` - Website ID from Umami
-
-### Frontend (.env)
-- `VITE_API_URL` - Backend API URL
-- `VITE_UMAMI_URL` - Umami analytics URL
-- `VITE_UMAMI_WEBSITE_ID` - Website ID from Umami
-- `VITE_UMAMI_SHARE_URL` - (Optional) Public share URL for embedded dashboard
-
-## Testing Approach
-- Jest with Supertest for API testing
-- Test files in `__tests__` directories
-- Database migrations run before tests
-- Mock email sending in tests
-
-## Umami Analytics Integration
-
-The frontend includes comprehensive Umami analytics integration for tracking user behavior and gallery performance.
-
-### Tracked Events:
-- **Gallery Events**:
-  - `gallery_password_entry` - Password attempts (success/failure)
-  - `gallery_photo_view` - Individual photo views
-  - `gallery_photo_download` - Single photo downloads
-  - `gallery_bulk_download` - Bulk/all photo downloads
-  - `gallery_expired` - Expired gallery access attempts
-- **Admin Events**:
-  - `admin_login` - Admin authentication
-  - `admin_event_created` - New event creation
-  - `admin_event_archived` - Event archiving
-  - `admin_event_deleted` - Event deletion
-  - `admin_settings_updated` - Settings changes
-- **User Behavior**:
-  - Search queries (with debouncing)
-  - Expiration warning views
-  - Page views with automatic tracking
-
-### Setup:
-1. Install Umami (self-hosted or cloud)
-2. Create a website in Umami dashboard
-3. Set environment variables:
-   ```
-   VITE_UMAMI_URL=https://your-umami-instance.com
-   VITE_UMAMI_WEBSITE_ID=your-website-id
-   VITE_UMAMI_SHARE_URL=https://your-umami-instance.com/share/...
-   ```
-
-### Analytics Dashboard:
-- Admin panel includes analytics page at `/admin/analytics`
-- Summary view with key metrics
-- Option to embed full Umami dashboard
-- Real-time event tracking
-
-## Accessibility & Performance Features
-
-### Accessibility (WCAG 2.1 AA Compliance)
-- **Error Boundaries**: Graceful error handling with recovery options
-- **Skip Links**: Skip to main content for keyboard navigation
-- **ARIA Labels**: Proper labeling for screen readers
-- **Focus Management**: Focus trap in modals, visible focus indicators
-- **Keyboard Navigation**: Full keyboard support in gallery lightbox (arrows, escape, +/-, d for download)
-- **Loading States**: Skeleton screens instead of spinners for better UX
-- **Offline Support**: Visual indicator when offline
-- **Form Validation**: Accessible error messages with aria-describedby
-
-### Performance Optimizations
-- **Lazy Loading**: Images load on scroll with Intersection Observer
-- **Skeleton Screens**: Instant visual feedback during loading
-- **Error Recovery**: Component-level error boundaries prevent full page crashes
-- **Optimistic Updates**: Immediate UI updates with background sync
-- **Debounced Search**: Prevents excessive API calls
-- **Analytics**: Non-blocking Umami integration
-
-### Component Library Enhancements
-- `<ErrorBoundary>` - Catches and displays errors gracefully
-- `<PageErrorBoundary>` - Full-page error recovery
-- `<Skeleton>` - Flexible skeleton loader with variants
-- `<OfflineIndicator>` - Network status monitoring
-- `<SkipLink>` - Accessibility navigation
-- `useFocusTrap` - Modal focus management hook
-- `useOnlineStatus` - Network status hook
-
-## Theme System & Branding
-
-### Theme Features
-- **Dynamic Theming**: CSS variables for runtime theme switching
-- **Preset Themes**: Default, Wedding, Birthday, Corporate, Minimal
-- **Customization Options**:
-  - Primary/Accent/Background/Text colors
-  - Font family selection
-  - Border radius (none, sm, md, lg)
-  - Custom logo upload
-  - Custom CSS injection
-- **Event-Specific Themes**: Override global theme per gallery
-- **Live Preview**: Real-time theme changes in admin panel
-
-### Theme Context API
-```typescript
-const { theme, setTheme, setThemeByName } = useTheme();
-```
-
-### Branding Settings
-- Company name, tagline, and support email
-- Custom footer text
-- Optional watermarking on downloads
-- Logo upload for gallery header
-
-### CSS Variables
-```css
---color-primary: #5C8762;
---color-primary-light: #7aa583;
---color-primary-dark: #4a6f4f;
---color-accent: #22c55e;
---color-background: #fafafa;
---color-text: #171717;
---font-family: 'Inter', sans-serif;
---border-radius: 0.5rem;
-```
-
-## Success Metrics (from PRD)
-- Time to generate gallery: <2 minutes
-- Guest satisfaction: >90%
-- System uptime: 99.9%
-- Email delivery rate: >98%
-- Successful archiving: 100%

CODE_OF_CONDUCT.md

@@ -0,0 +1,27 @@
+# PicPeak Community Guidelines
+
+## Our Commitment
+
+We are committed to providing a welcoming and inspiring community for all photographers and developers.
+
+## Expected Behavior
+
+* Be respectful and considerate
+* Welcome newcomers and help them get started
+* Focus on what is best for the community
+* Show empathy towards other community members
+
+## Unacceptable Behavior
+
+* Trolling or insulting comments
+* Personal attacks
+* Public or private harassment
+* Publishing others' private information
+
+## Enforcement
+
+Instances of unacceptable behavior may be reported by [opening an issue](https://github.com/PicPeak/picpeak/issues/new?labels=conduct) on GitHub. All complaints will be reviewed and investigated promptly and fairly.
+
+## Attribution
+
+This Code of Conduct is adapted from contributor-covenant.org, version 2.0.

CONTRIBUTING.md

@@ -0,0 +1,193 @@
+# Contributing to PicPeak
+
+First off, thank you for considering contributing to PicPeak! It's people like you that make PicPeak such a great tool for photographers worldwide.
+
+## 🤝 Code of Conduct
+
+This project and everyone participating in it is governed by the [PicPeak Code of Conduct](CODE_OF_CONDUCT.md). By participating, you are expected to uphold this code.
+
+## 🎯 How Can I Contribute?
+
+### Reporting Bugs
+
+Before creating bug reports, please check the existing issues as you might find out that you don't need to create one. When you are creating a bug report, please include as many details as possible:
+
+* **Use a clear and descriptive title**
+* **Describe the exact steps to reproduce the problem**
+* **Provide specific examples to demonstrate the steps**
+* **Describe the behavior you observed and what you expected**
+* **Include screenshots if possible**
+* **Include your environment details** (OS, browser, Docker version, etc.)
+
+### Suggesting Enhancements
+
+Enhancement suggestions are tracked as GitHub issues. When creating an enhancement suggestion, please include:
+
+* **Use a clear and descriptive title**
+* **Provide a detailed description of the suggested enhancement**
+* **Provide specific examples to demonstrate the enhancement**
+* **Describe the current behavior and expected behavior**
+* **Explain why this enhancement would be useful**
+
+### Your First Code Contribution
+
+Unsure where to begin? You can start by looking through these issues:
+
+* [Good first issues](https://github.com/PicPeak/picpeak/labels/good%20first%20issue) - issues which should only require a few lines of code
+* [Help wanted issues](https://github.com/PicPeak/picpeak/labels/help%20wanted) - issues which need extra attention
+
+### Pull Requests
+
+1. **Fork the repo** and create your branch from `main` (active development)
+2. **Install dependencies**:
+   ```bash
+   cd backend && npm install
+   cd ../frontend && npm install
+   ```
+3. **Make your changes** and ensure:
+   - Code follows the existing style
+   - Tests pass: `npm test`
+   - Linting passes: `npm run lint`
+4. **Write tests** if you've added code
+5. **Update documentation** if needed
+6. **Attach a screenshot for any UI change** (see below)
+7. **Create a Pull Request**
+
+> **📸 Screenshots are required for UI changes.** Any PR that changes a user-facing surface — a component, page, layout, style, or in-app copy — must include at least one screenshot of the result in the PR description, showing before/after where it helps reviewers see the difference. PRs that touch the UI without a screenshot will be asked to add one before review. Backend-only or otherwise non-visual changes don't need one.
+
+## 💻 Development Setup
+
+### Prerequisites
+
+- Node.js 18+
+- Docker & Docker Compose
+- Git
+
+### Local Development
+
+```bash
+# Clone your fork
+git clone https://github.com/your-username/picpeak.git
+cd picpeak
+
+# Install dependencies
+cd backend && npm install
+cd ../frontend && npm install
+
+# Set up environment
+cp .env.example .env
+# Edit .env with your settings
+
+# Start development servers
+docker-compose -f docker-compose.dev.yml up
+```
+
+**After pulling changes that touch `backend/package.json` / `backend/package-lock.json` (or the frontend equivalents)**, rebuild the affected image so the live-mounted source can `require()` the new deps:
+
+```bash
+docker compose -f docker-compose.dev.yml up -d --build backend
+# (or `frontend`, or both)
+```
+
+The dev compose bakes `node_modules` into the image while live-mounting `./backend/src` and `./frontend/src` from disk. A dep added on disk won't be picked up until the image is rebuilt — typical symptom is a `MODULE_NOT_FOUND` restart loop on the affected container.
+
+### Running Tests
+
+```bash
+# Backend tests
+cd backend && npm test
+
+# Frontend tests
+cd frontend && npm test
+
+# E2E tests
+npm run test:e2e
+```
+
+## 📝 Styleguides
+
+### Git Commit Messages
+
+* Use the present tense ("Add feature" not "Added feature")
+* Use the imperative mood ("Move cursor to..." not "Moves cursor to...")
+* Limit the first line to 72 characters or less
+* Reference issues and pull requests liberally after the first line
+* Consider starting the commit message with an applicable emoji:
+  * 🎨 `:art:` when improving the format/structure of the code
+  * 🐛 `:bug:` when fixing a bug
+  * 🔥 `:fire:` when removing code or files
+  * 📝 `:memo:` when writing docs
+  * 🚀 `:rocket:` when improving performance
+  * ✨ `:sparkles:` when adding a new feature
+
+### JavaScript/TypeScript Styleguide
+
+* Use ES6+ features
+* Prefer async/await over promises
+* Use meaningful variable names
+* Add JSDoc comments for functions
+* Follow ESLint rules
+
+### React Styleguide
+
+* Use functional components with hooks
+* Keep components small and focused
+* Use TypeScript for type safety
+* Follow the existing folder structure
+* Write tests for new components
+
+## 📦 Project Structure
+
+```
+picpeak/
+├── backend/
+│   ├── src/
+│   │   ├── routes/      # API endpoints
+│   │   ├── services/    # Business logic
+│   │   ├── middleware/  # Express middleware
+│   │   └── utils/       # Utilities
+│   └── migrations/      # Database migrations
+├── frontend/
+│   ├── src/
+│   │   ├── components/  # Reusable components
+│   │   ├── pages/       # Page components
+│   │   ├── services/    # API services
+│   │   └── hooks/       # Custom hooks
+│   └── public/          # Static assets
+```
+
+## 🌿 Branch model
+
+PicPeak runs on two long-lived branches:
+
+| Branch | Role | What targets it |
+|---|---|---|
+| **`main`** | Active development. The next release is being assembled here. | Feature PRs. Most bugfix PRs. |
+| **`stable`** | Curated release channel. Production-recommended. | Security fixes and regular bugfix backports, kept small and free of unrelated features. |
+
+### Which branch should my PR target?
+
+- **New feature** → target `main`.
+- **Bugfix that ONLY affects active dev** → target `main`.
+- **Bugfix that current stable users need** → target `main`; regular bug fixes are generally backported automatically to `stable`. Maintainers handle conflicts or create a separate focused backport PR when needed.
+- **Security vulnerability** → report privately using [SECURITY.md](SECURITY.md). Security fixes are always released on both `stable` and `main`; coordinate any fix with the maintainers before opening a public PR.
+
+**Hard rule on PR scope**: bugfix PRs against `stable` must be small enough to backport without conflict. Omnibus PRs (e.g. five unrelated sub-features) are fine for `main`, but never for `stable` — they make the next `main → stable` merge painful and break the "stable is always shippable" invariant.
+
+If you're not sure which branch to target, default to `main` and a maintainer will retarget during review.
+
+## 🔄 Release Process
+
+Releases are cut independently from `main` (pre-release versions for the active channel) and `stable` (semver releases for the curated channel). `release-please` handles version bumps, changelog generation, and Docker image publication automatically — contributors don't update `package.json` or `CHANGELOG.md` by hand.
+
+Periodic `main → stable` merges promote a batch of `main` work to the stable channel. The maintainer chooses when (typically every ~4 weeks, sooner if a hot bug demands it).
+
+See [RELEASING.md](RELEASING.md) for the full operational doc (promotion criteria, conflict-resolution checklist for the `main → stable` merge, hotfix backport path, versioning rules).
+
+## 📮 Contact
+
+- Create an [issue](https://github.com/PicPeak/picpeak/issues) for bugs or features
+- Join [discussions](https://github.com/PicPeak/picpeak/discussions) for questions
+- Security vulnerabilities: Follow the [security policy](SECURITY.md) and use [private vulnerability reporting](https://github.com/PicPeak/picpeak/security/advisories/new)
+
+Thank you for contributing! 🎉

DEPLOYMENT-TRAEFIK.md

@@ -1,92 +0,0 @@
-# Deployment Guide - Traefik Production Setup
-
-## Overview
-
-This guide explains how to deploy PicPeak with an external Traefik reverse proxy for production use.
-
-## Fixed Issues
-
-1. **Database Migration**: Added missing `created_at` column to `email_queue` table
-2. **502 Bad Gateway**: Properly configured Traefik routing and backend accessibility
-3. **Health Checks**: Fixed health check endpoint imports and paths
-
-## Deployment Steps
-
-### 1. Update Environment Variables
-
-Ensure your `.env` file has the correct URLs:
-```bash
-ADMIN_URL=https://picpeak.nothaft.cloud
-FRONTEND_URL=https://picpeak.nothaft.cloud
-```
-
-### 2. Build Images
-
-```bash
-# Build backend image
-docker build -t picpeak-backend:latest ./backend
-
-# Build frontend image  
-docker build -t picpeak-frontend:latest ./frontend \
-  --build-arg VITE_API_URL=/api \
-  --build-arg VITE_UMAMI_URL=${VITE_UMAMI_URL} \
-  --build-arg VITE_UMAMI_WEBSITE_ID=${VITE_UMAMI_WEBSITE_ID}
-```
-
-### 3. Deploy with Traefik
-
-Use the new Traefik-specific compose file:
-```bash
-docker-compose -f docker-compose.traefik.yml up -d
-```
-
-### 4. Verify Deployment
-
-Check that all services are healthy:
-```bash
-# Check container status
-docker-compose -f docker-compose.traefik.yml ps
-
-# Check backend health
-curl https://picpeak.nothaft.cloud/api/health
-
-# Check logs
-docker-compose -f docker-compose.traefik.yml logs -f backend
-```
-
-## Key Differences from Standard Deployment
-
-1. **No Internal Nginx**: Traefik handles all routing externally
-2. **API Path Stripping**: Traefik strips `/api` prefix when forwarding to backend
-3. **Network Configuration**: Services join external `traefik` network
-4. **Health Checks**: Backend exposes `/health` endpoint (not `/api/health`)
-
-## Why CI/CD Tests Pass But Production Fails
-
-CI/CD tests typically:
-- Use in-memory or temporary databases with fresh migrations
-- Don't test through reverse proxy (direct API calls)
-- Don't run background services (email processor, etc.)
-- Have different network configurations
-
-Production environment has:
-- Persistent database that may have migration state issues
-- Reverse proxy routing complexity
-- All background services running
-- Different security and network constraints
-
-## Troubleshooting
-
-### 502 Bad Gateway
-- Check Traefik network connectivity: `docker network ls`
-- Verify backend is in traefik network: `docker inspect picpeak-backend`
-- Check Traefik logs: `docker logs traefik`
-
-### Database Issues
-- Connect to database: `docker exec -it picpeak-db psql -U picpeak`
-- Check migration status: `SELECT * FROM migrations;`
-- Run migrations manually: `docker exec -it picpeak-backend npm run migrate:safe`
-
-### Email Service Errors
-- Check email queue: `SELECT * FROM email_queue ORDER BY created_at DESC LIMIT 10;`
-- Monitor email processor: `docker logs picpeak-backend | grep "email"`

DEPLOYMENT.md

@@ -1,346 +0,0 @@
-# PicPeak Deployment Guide
-
-This guide covers deploying PicPeak for development and production environments.
-
-## Table of Contents
-- [Quick Start (Development)](#quick-start-development)
-- [Production Deployment](#production-deployment)
-- [Admin User Setup](#admin-user-setup)
-- [Configuration Reference](#configuration-reference)
-- [Troubleshooting](#troubleshooting)
-
-## Quick Start (Development)
-
-### 1. Clone and Setup
-
-```bash
-git clone https://github.com/yourusername/picpeak.git
-cd picpeak
-
-# Copy environment template
-cp .env.example .env
-
-# Start development environment
-docker-compose -f docker-compose.dev.yml up -d
-```
-
-### 2. Access Services
-
-- Frontend: http://localhost:3005
-- Backend API: http://localhost:3001
-- MailHog (email testing): http://localhost:8025
-
-### 3. Create Admin User
-
-```bash
-docker-compose -f docker-compose.dev.yml exec backend node scripts/create-admin.js \
-  --email admin@localhost \
-  --username admin \
-  --password admin123
-```
-
-## Production Deployment
-
-### Prerequisites
-
-- Docker and Docker Compose installed
-- Domain with DNS configured
-- SSL/TLS handled by reverse proxy (Traefik, Nginx, etc.)
-
-### 1. Environment Setup
-
-```bash
-# Copy production template
-cp .env.production.example .env
-
-# Generate secure secrets
-echo "JWT_SECRET=$(openssl rand -base64 32)" >> .env
-echo "DB_PASSWORD=$(openssl rand -base64 24)" >> .env
-```
-
-Edit `.env` with your configuration:
-
-```env
-# Your domain
-ADMIN_URL=https://yourdomain.com
-FRONTEND_URL=https://yourdomain.com
-
-# Database (PostgreSQL)
-DB_USER=picpeak
-DB_NAME=picpeak
-# DB_PASSWORD already generated above
-
-# Email
-SMTP_HOST=smtp.gmail.com
-SMTP_PORT=587
-SMTP_SECURE=true
-SMTP_USER=your-email@gmail.com
-SMTP_PASS=your-app-password
-EMAIL_FROM=noreply@yourdomain.com
-```
-
-### 2. Frontend Configuration
-
-```bash
-# Configure frontend for production
-echo "VITE_API_URL=/api" > frontend/.env.production
-```
-
-### 3. Deploy with Docker Compose
-
-```bash
-# Build and start services
-docker-compose -f docker-compose.prod.yml up -d
-
-# Check status
-docker-compose -f docker-compose.prod.yml ps
-
-# View logs
-docker-compose -f docker-compose.prod.yml logs -f
-```
-
-### 4. Deploy with Traefik
-
-If using Traefik, create `docker-compose.override.yml`:
-
-```yaml
-version: '3.8'
-
-services:
-  frontend:
-    labels:
-      - "traefik.enable=true"
-      - "traefik.http.routers.picpeak.rule=Host(`yourdomain.com`)"
-      - "traefik.http.routers.picpeak.entrypoints=websecure"
-      - "traefik.http.routers.picpeak.tls.certresolver=letsencrypt"
-      - "traefik.http.services.picpeak.loadbalancer.server.port=80"
-    networks:
-      - traefik
-      - picpeak
-
-networks:
-  traefik:
-    external: true
-```
-
-## Admin User Setup
-
-### Create First Admin
-
-After deployment, create your admin user:
-
-```bash
-# Production
-docker-compose -f docker-compose.prod.yml exec backend node scripts/create-admin.js \
-  --email admin@yourdomain.com \
-  --username admin \
-  --password yourSecurePassword
-
-# Auto-generate password
-docker-compose -f docker-compose.prod.yml exec backend node scripts/create-admin.js \
-  --email admin@yourdomain.com
-```
-
-The script will display:
-- ✅ Admin user created successfully!
-- Email: admin@yourdomain.com
-- Username: admin
-- Login URL: https://yourdomain.com/admin/login
-- Password: (save this if auto-generated!)
-
-### Managing Admin Users
-
-```bash
-# List admin users
-docker-compose -f docker-compose.prod.yml exec backend \
-  psql postgresql://picpeak:$DB_PASSWORD@db:5432/picpeak \
-  -c "SELECT id, username, email, is_active, last_login FROM admin_users;"
-
-# Deactivate user
-docker-compose -f docker-compose.prod.yml exec backend \
-  psql postgresql://picpeak:$DB_PASSWORD@db:5432/picpeak \
-  -c "UPDATE admin_users SET is_active = false WHERE email = 'user@example.com';"
-```
-
-## Configuration Reference
-
-### Database Configuration
-
-PicPeak automatically detects the environment and uses:
-- **Development**: SQLite (`./data/photo_sharing.db`)
-- **Production**: PostgreSQL (configured via environment variables)
-
-### Environment Variables
-
-#### Required for Production
-
-| Variable | Description | Example |
-|----------|-------------|---------|
-| `JWT_SECRET` | JWT signing key | `openssl rand -base64 32` |
-| `DB_PASSWORD` | PostgreSQL password | `openssl rand -base64 24` |
-| `ADMIN_URL` | Admin panel URL | `https://yourdomain.com` |
-| `FRONTEND_URL` | Frontend URL | `https://yourdomain.com` |
-| `EMAIL_FROM` | Sender email | `noreply@yourdomain.com` |
-
-#### Email Configuration
-
-| Variable | Description | Example |
-|----------|-------------|---------|
-| `SMTP_HOST` | SMTP server | `smtp.gmail.com` |
-| `SMTP_PORT` | SMTP port | `587` |
-| `SMTP_SECURE` | Use TLS | `true` |
-| `SMTP_USER` | SMTP username | `your-email@gmail.com` |
-| `SMTP_PASS` | SMTP password | App-specific password |
-
-### Storage Paths
-
-- Photos: `./storage/events/active/`
-- Archives: `./storage/events/archived/`
-- Thumbnails: `./storage/thumbnails/`
-- Uploads: `./storage/uploads/`
-
-## Backup and Restore
-
-### Backup Database
-
-```bash
-# PostgreSQL backup
-docker-compose -f docker-compose.prod.yml exec db \
-  pg_dump -U picpeak picpeak > backup-$(date +%Y%m%d).sql
-
-# Backup storage
-tar -czf storage-backup-$(date +%Y%m%d).tar.gz ./storage
-```
-
-### Restore Database
-
-```bash
-# PostgreSQL restore
-docker-compose -f docker-compose.prod.yml exec -T db \
-  psql -U picpeak picpeak < backup-20240115.sql
-
-# Restore storage
-tar -xzf storage-backup-20240115.tar.gz
-```
-
-## Monitoring
-
-### Health Checks
-
-```bash
-# Backend health
-curl https://yourdomain.com/api/health
-
-# Frontend health
-curl https://yourdomain.com/health
-```
-
-### Logs
-
-```bash
-# All services
-docker-compose -f docker-compose.prod.yml logs -f
-
-# Specific service
-docker-compose -f docker-compose.prod.yml logs -f backend
-
-# Last 100 lines
-docker-compose -f docker-compose.prod.yml logs --tail=100 backend
-```
-
-## Troubleshooting
-
-### Backend Won't Start
-
-1. Check database connection:
-   ```bash
-   docker-compose -f docker-compose.prod.yml logs db
-   ```
-
-2. Verify environment variables:
-   ```bash
-   docker-compose -f docker-compose.prod.yml exec backend env | grep DB_
-   ```
-
-### Can't Login as Admin
-
-1. Verify admin user exists:
-   ```bash
-   docker-compose -f docker-compose.prod.yml exec backend \
-     psql postgresql://picpeak:$DB_PASSWORD@db:5432/picpeak \
-     -c "SELECT * FROM admin_users;"
-   ```
-
-2. Reset admin password:
-   ```bash
-   # Create new admin with different email
-   docker-compose -f docker-compose.prod.yml exec backend \
-     node scripts/create-admin.js --email newadmin@yourdomain.com
-   ```
-
-### Photos Not Loading
-
-1. Check file permissions:
-   ```bash
-   ls -la ./storage/events/active/
-   ```
-
-2. Verify nginx proxy configuration:
-   ```bash
-   docker-compose -f docker-compose.prod.yml exec frontend \
-     cat /etc/nginx/conf.d/default.conf
-   ```
-
-### Email Not Sending
-
-1. Check email configuration:
-   ```bash
-   docker-compose -f docker-compose.prod.yml exec backend env | grep SMTP_
-   ```
-
-2. View email queue:
-   ```bash
-   docker-compose -f docker-compose.prod.yml exec backend \
-     psql postgresql://picpeak:$DB_PASSWORD@db:5432/picpeak \
-     -c "SELECT * FROM email_queue WHERE status = 'failed';"
-   ```
-
-## Maintenance
-
-### Update Application
-
-```bash
-# Pull latest changes
-git pull
-
-# Rebuild images
-docker-compose -f docker-compose.prod.yml build
-
-# Restart services
-docker-compose -f docker-compose.prod.yml up -d
-```
-
-### Clean Up
-
-```bash
-# Remove unused images
-docker image prune -a
-
-# Clean up logs
-docker-compose -f docker-compose.prod.yml logs --tail=0 -f
-
-# Remove old archives
-find ./storage/events/archived -name "*.zip" -mtime +90 -delete
-```
-
-## Security Checklist
-
-- [ ] Generated secure `JWT_SECRET`
-- [ ] Generated secure `DB_PASSWORD`
-- [ ] HTTPS enabled via reverse proxy
-- [ ] Changed default admin credentials
-- [ ] Configured real SMTP server
-- [ ] Set file permissions: `chmod 600 .env`
-- [ ] Firewall configured
-- [ ] Regular backups scheduled
-- [ ] Monitoring enabled

MIGRATION_ERROR_FIX.md

@@ -1,111 +0,0 @@
-# Quick Fix for Migration Error
-
-## Immediate Fix
-
-The error "relation photo_categories already exists" occurs because the database already has tables but the migration tracking doesn't know they were applied.
-
-### Option 1: Use Safe Migration Runner (Recommended)
-
-Update your `docker-compose.prod.yml` to use the safe migration command:
-
-```yaml
-backend:
-  environment:
-    - NODE_ENV=production
-  # ... other config ...
-```
-
-Then update the `wait-for-db.sh` (already done) to use `npm run migrate:safe` in production.
-
-### Option 2: Quick Manual Fix
-
-If you need to fix the running system immediately:
-
-```bash
-# 1. Enter the backend container
-docker-compose -f docker-compose.prod.yml exec backend sh
-
-# 2. Run the safe migration script
-npm run migrate:safe
-
-# 3. If that fails, manually mark migrations as applied:
-docker-compose -f docker-compose.prod.yml exec db psql -U picpeak -d picpeak
-
-# In PostgreSQL:
-CREATE TABLE IF NOT EXISTS migrations (
-  id SERIAL PRIMARY KEY,
-  filename VARCHAR(255) UNIQUE NOT NULL,
-  applied_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
-);
-
--- Mark existing migrations as applied
-INSERT INTO migrations (filename) VALUES 
-  ('init.js'),
-  ('004_add_categories_and_cms.js'),
-  ('006_add_photo_counter_to_categories.js'),
-  ('007_add_read_at_to_activity_logs.js'),
-  ('008_add_language_support_to_email_templates.js'),
-  ('009_update_german_email_templates.js'),
-  ('010_add_missing_email_templates.js'),
-  ('011_add_user_upload_settings.js'),
-  ('012_add_hero_photo_id.js'),
-  ('013_fix_email_links_and_date_format.js'),
-  ('014_add_default_welcome_message.js'),
-  ('014_add_host_name_to_events.js'),
-  ('015_add_login_attempts_table.js'),
-  ('016_add_auth_security_columns.js'),
-  ('017_add_token_revocation_tables.js')
-ON CONFLICT (filename) DO NOTHING;
-
-\q
-```
-
-### Option 3: Fresh Start (Nuclear Option)
-
-If you don't have important data yet:
-
-```bash
-# Stop everything
-docker-compose -f docker-compose.prod.yml down
-
-# Remove database volume
-docker volume rm wedding-photo-sharing_postgres_data
-
-# Start fresh
-docker-compose -f docker-compose.prod.yml up -d
-```
-
-## Root Cause
-
-The issue happens when:
-1. Database volume persists between deployments
-2. Migration tracking table gets out of sync
-3. The original migration runner doesn't check for existing tables
-
-## Permanent Solution
-
-The new safe migration runner (`migrate:safe`) handles this by:
-1. Checking if tables exist before creating them
-2. Catching "already exists" errors gracefully
-3. Auto-detecting existing schema and marking migrations as applied
-
-## Next Steps
-
-After fixing the migration issue:
-
-1. Create admin user:
-```bash
-docker-compose -f docker-compose.prod.yml exec backend node scripts/create-admin.js \
-  --username admin \
-  --email admin@yourdomain.com
-```
-
-2. Check health:
-```bash
-curl http://yourdomain.com/api/health
-```
-
-3. Monitor logs:
-```bash
-docker-compose -f docker-compose.prod.yml logs -f backend
-```

PRODUCTION_DEPLOYMENT.md

@@ -1,98 +0,0 @@
-# Production Deployment Guide
-
-This guide explains how to deploy PicPeak in production behind a reverse proxy like Traefik.
-
-## Environment Configuration
-
-### Frontend Configuration
-
-For production deployment behind a reverse proxy (Traefik, Nginx, etc.), the frontend should use relative URLs to automatically inherit the protocol (HTTPS) and domain.
-
-1. Copy the production environment template:
-   ```bash
-   cp frontend/.env.production.example frontend/.env.production
-   ```
-
-2. Set the API URL to use relative path:
-   ```env
-   # frontend/.env.production
-   VITE_API_URL=/api
-   ```
-
-   This ensures all API calls will use the same domain and protocol as the frontend.
-
-### Backend Configuration
-
-Ensure your backend `.env` file has the correct URLs:
-```env
-# backend/.env
-FRONTEND_URL=https://yourdomain.com
-ADMIN_URL=https://yourdomain.com
-```
-
-## Docker Compose Production
-
-When using Docker Compose in production:
-
-1. Build with production environment:
-   ```bash
-   docker-compose -f docker-compose.prod.yml build --build-arg NODE_ENV=production
-   ```
-
-2. The frontend nginx configuration already includes proper proxy settings for:
-   - `/api` → Backend API
-   - `/photos` → Protected photo access
-   - `/thumbnails` → Thumbnail images
-   - `/uploads` → Public uploads (logos, favicons)
-
-## Traefik Configuration
-
-Example Traefik labels for docker-compose:
-
-```yaml
-services:
-  frontend:
-    labels:
-      - "traefik.enable=true"
-      - "traefik.http.routers.picpeak.rule=Host(`yourdomain.com`)"
-      - "traefik.http.routers.picpeak.entrypoints=websecure"
-      - "traefik.http.routers.picpeak.tls.certresolver=letsencrypt"
-      - "traefik.http.services.picpeak.loadbalancer.server.port=80"
-```
-
-## Important Notes
-
-1. **No Hardcoded URLs**: The application uses environment variables with relative URL fallbacks, making it production-ready.
-
-2. **HTTPS Only**: When `VITE_API_URL=/api`, all requests will use the same protocol as the page (HTTPS in production).
-
-3. **CORS Configuration**: The backend CORS is configured to accept requests from the URLs specified in `FRONTEND_URL` and `ADMIN_URL`.
-
-4. **Static Assets**: All static assets (photos, thumbnails, uploads) are served through the nginx proxy, inheriting authentication headers.
-
-## Verification
-
-After deployment, verify:
-
-1. Check browser console for any localhost URLs (there should be none)
-2. Verify all API calls use HTTPS
-3. Check that images load correctly with authentication
-4. Test favicon and logo display
-
-## Troubleshooting
-
-If you see console errors about localhost:
-
-1. Ensure `VITE_API_URL=/api` in frontend environment
-2. Clear browser cache
-3. Rebuild frontend with production environment:
-   ```bash
-   cd frontend
-   npm run build
-   ```
-
-If images don't load:
-
-1. Check that nginx proxy locations are configured
-2. Verify authentication tokens are being sent
-3. Check backend logs for authentication errors

PRODUCTION_DEPLOYMENT_FIXES.md

@@ -1,100 +0,0 @@
-# Production Deployment Fixes
-
-This document describes the fixes applied to resolve production deployment issues in Docker.
-
-## Issues Fixed
-
-### 1. Database Connection Error: "getaddrinfo ENOTFOUND postgres"
-**Problem**: The backend was trying to connect to hostname "postgres" but the database service is named "db" in docker-compose.
-**Solution**: 
-- Updated `knexfile.js` to use correct default host "db" instead of "postgres"
-- Added `depends_on: db` to backend service in docker-compose.prod.yml
-
-### 2. Backend Starting Before Database Ready
-**Problem**: Backend service started before PostgreSQL was ready, causing connection failures.
-**Solution**:
-- Created `wait-for-db.sh` script that waits for PostgreSQL to be ready
-- Updated Dockerfile to install postgresql-client and use the wait script
-- Script also runs migrations automatically on startup
-
-### 3. Email Processor Initialization Failure
-**Problem**: Email processor tried to initialize on module load before database was available.
-**Solution**:
-- Modified `emailProcessor.js` to export initialization functions
-- Updated `server.js` to call initialization after database is ready
-- Added proper error handling for email service initialization
-
-### 4. Missing Environment Variables
-**Problem**: Critical storage path environment variables were missing.
-**Solution**:
-- Added STORAGE_PATH, EVENTS_PATH, and ARCHIVE_PATH to docker-compose.prod.yml
-- Created `.env.example` documenting all required environment variables
-
-### 5. Enhanced Health Check
-**Problem**: Basic health check didn't verify database connectivity.
-**Solution**:
-- Updated `/api/health` endpoint to check database connection
-- Returns proper HTTP 503 status when unhealthy
-
-## Files Modified
-
-1. **backend/knexfile.js** - Fixed production database defaults
-2. **backend/wait-for-db.sh** - Created database wait script
-3. **backend/Dockerfile** - Added postgresql-client and wait script
-4. **docker-compose.prod.yml** - Added dependencies and environment variables
-5. **backend/src/services/emailProcessor.js** - Disabled auto-initialization
-6. **backend/server.js** - Added email initialization and improved health check
-7. **backend/.env.example** - Created environment variable documentation
-
-## Deployment Steps
-
-1. Ensure all environment variables are set according to `.env.example`
-2. Build and deploy with docker-compose:
-   ```bash
-   docker-compose -f docker-compose.prod.yml build
-   docker-compose -f docker-compose.prod.yml up -d
-   ```
-3. The backend will now:
-   - Wait for PostgreSQL to be ready
-   - Run migrations automatically
-   - Initialize all services in proper order
-   - Provide health status at `/api/health`
-
-## Verification
-
-Check deployment health:
-```bash
-curl http://localhost/api/health
-```
-
-Expected response:
-```json
-{
-  "status": "ok",
-  "database": "connected",
-  "timestamp": "2025-07-13T20:30:00.000Z"
-}
-```
-
-## Email Configuration
-
-Email service requires configuration in the database. If email is not configured:
-- The service will log a warning but continue running
-- Emails will be queued but not sent
-- Configure email settings in the admin panel after deployment
-
-## PostgreSQL Connection Fix
-
-### Issue: "no pg_hba.conf entry for host"
-This error occurs when PostgreSQL requires SSL but the client connects without encryption.
-
-### Solution:
-- Disabled SSL requirement for PostgreSQL in Docker environment (`ssl=off`)
-- Added proper authentication method (`scram-sha-256`)
-- This is acceptable for internal Docker networks where all traffic is isolated
-
-### Security Note:
-For production deployments exposed to the internet:
-1. Use SSL certificates for PostgreSQL
-2. Or ensure the database is only accessible within the Docker network
-3. Never expose PostgreSQL port (5432) directly to the internet

PRODUCTION_DEPLOYMENT_GUIDE.md

@@ -1,312 +0,0 @@
-# Production Deployment Guide
-
-This guide addresses all known production deployment issues and provides solutions.
-
-## Pre-Deployment Checklist
-
-### 1. Environment Variables
-Create a `.env` file with ALL required variables:
-
-```bash
-# Required
-JWT_SECRET=<generate-with-openssl-rand-base64-32>
-DB_PASSWORD=<strong-password>
-ADMIN_URL=https://yourdomain.com
-FRONTEND_URL=https://yourdomain.com
-
-# Database
-DB_USER=picpeak
-DB_NAME=picpeak
-
-# Email (Optional but recommended)
-SMTP_HOST=smtp.gmail.com
-SMTP_PORT=587
-SMTP_SECURE=false
-SMTP_USER=your-email@gmail.com
-SMTP_PASS=your-app-password
-EMAIL_FROM=noreply@yourdomain.com
-
-# Umami Analytics (Optional)
-UMAMI_URL=https://analytics.yourdomain.com
-UMAMI_WEBSITE_ID=your-website-id
-UMAMI_HASH_SALT=<generate-random-string>
-```
-
-### 2. Generate Secrets
-
-```bash
-# Generate JWT Secret
-openssl rand -base64 32
-
-# Generate Database Password
-openssl rand -base64 24
-
-# Generate Umami Hash Salt
-openssl rand -hex 32
-```
-
-## Deployment Steps
-
-### 1. Initial Setup
-
-```bash
-# Clone repository
-git clone https://github.com/yourusername/wedding-photo-sharing.git
-cd wedding-photo-sharing
-
-# Create required directories
-mkdir -p storage/events/active storage/events/archived storage/thumbnails storage/uploads
-mkdir -p data logs
-mkdir -p certbot/conf certbot/www
-
-# Set permissions (important!)
-chmod -R 755 storage data logs
-```
-
-### 2. Fix Docker Volume Permissions
-
-Create `docker-compose.override.yml` for local volume configuration:
-
-```yaml
-version: '3.8'
-
-services:
-  backend:
-    volumes:
-      - ./storage:/app/storage:delegated
-      - ./data:/app/data:delegated
-      - ./logs:/app/logs:delegated
-    user: "1001:1001"  # nodejs user
-
-  db:
-    volumes:
-      - ./postgres-data:/var/lib/postgresql/data
-```
-
-### 3. Build and Deploy
-
-```bash
-# Build images
-docker-compose -f docker-compose.prod.yml build
-
-# Start services
-docker-compose -f docker-compose.prod.yml up -d
-
-# Check logs
-docker-compose -f docker-compose.prod.yml logs -f backend
-```
-
-### 4. Create Admin User
-
-After deployment, create the first admin user:
-
-```bash
-# Enter backend container
-docker-compose -f docker-compose.prod.yml exec backend sh
-
-# Create admin
-node scripts/create-admin.js \
-  --username admin \
-  --email admin@yourdomain.com \
-  --password <your-secure-password>
-
-# Exit container
-exit
-```
-
-### 5. Configure Email (if using database config)
-
-1. Login to admin panel: https://yourdomain.com/admin
-2. Go to Settings > Email Configuration
-3. Enter SMTP details
-4. Test email sending
-
-## Common Issues and Solutions
-
-### Issue 1: Migration Failures
-
-**Error**: "relation already exists"
-
-**Solution**: The safe migration runner handles this automatically. If issues persist:
-
-```bash
-# Reset migrations tracking
-docker-compose -f docker-compose.prod.yml exec db psql -U picpeak -d picpeak
-
-# In PostgreSQL:
-DROP TABLE IF EXISTS migrations;
-\q
-
-# Re-run migrations
-docker-compose -f docker-compose.prod.yml exec backend npm run migrate:safe
-```
-
-### Issue 2: Permission Denied Errors
-
-**Error**: "EACCES: permission denied"
-
-**Solution**: Fix container permissions:
-
-```bash
-# Stop containers
-docker-compose -f docker-compose.prod.yml down
-
-# Fix permissions on host
-sudo chown -R 1001:1001 storage data logs
-
-# Restart
-docker-compose -f docker-compose.prod.yml up -d
-```
-
-### Issue 3: Database Connection Failed
-
-**Error**: "no pg_hba.conf entry"
-
-**Solution**: Already fixed in docker-compose.prod.yml with:
-- SSL disabled for internal Docker network
-- Proper authentication method (scram-sha-256)
-
-### Issue 4: Frontend Can't Connect to Backend
-
-**Error**: CORS errors or connection refused
-
-**Solution**: Ensure environment variables match:
-- Backend: `FRONTEND_URL` must match your frontend URL
-- Frontend: `VITE_API_URL` must be set during build
-
-### Issue 5: Email Not Sending
-
-**Solution**: Check email configuration:
-
-```bash
-# Check backend logs
-docker-compose -f docker-compose.prod.yml logs backend | grep email
-
-# Verify SMTP settings
-# Gmail users: Use app password, not regular password
-# Enable "Less secure app access" or use OAuth2
-```
-
-## SSL/HTTPS Setup
-
-1. Update `nginx/sites-enabled/default` with your domain
-2. Run certbot:
-
-```bash
-# Initial certificate
-docker-compose -f docker-compose.prod.yml run --rm certbot certonly \
-  --webroot --webroot-path=/var/www/certbot \
-  -d yourdomain.com -d www.yourdomain.com
-
-# Auto-renewal is handled by the certbot container
-```
-
-## Monitoring
-
-### Health Checks
-
-```bash
-# Backend health
-curl http://localhost/api/health
-
-# Database connection
-docker-compose -f docker-compose.prod.yml exec backend \
-  psql -U picpeak -d picpeak -c "SELECT 1"
-```
-
-### Logs
-
-```bash
-# All services
-docker-compose -f docker-compose.prod.yml logs -f
-
-# Specific service
-docker-compose -f docker-compose.prod.yml logs -f backend
-```
-
-## Backup and Restore
-
-### Backup
-
-```bash
-#!/bin/bash
-# backup.sh
-DATE=$(date +%Y%m%d_%H%M%S)
-BACKUP_DIR="./backups/$DATE"
-
-mkdir -p $BACKUP_DIR
-
-# Database
-docker-compose -f docker-compose.prod.yml exec -T db \
-  pg_dump -U picpeak picpeak > $BACKUP_DIR/database.sql
-
-# Files
-tar -czf $BACKUP_DIR/storage.tar.gz storage/
-
-echo "Backup completed: $BACKUP_DIR"
-```
-
-### Restore
-
-```bash
-# Database
-docker-compose -f docker-compose.prod.yml exec -T db \
-  psql -U picpeak picpeak < ./backups/20240713_120000/database.sql
-
-# Files
-tar -xzf ./backups/20240713_120000/storage.tar.gz
-```
-
-## Production Best Practices
-
-1. **Always use named volumes** in production for better data persistence
-2. **Set up monitoring** with Prometheus/Grafana
-3. **Enable backups** with automated scripts
-4. **Use a reverse proxy** (Nginx) for SSL termination
-5. **Implement rate limiting** at the Nginx level
-6. **Regular updates** - Keep Docker images updated
-7. **Log rotation** - Configure log rotation for application logs
-
-## Troubleshooting Commands
-
-```bash
-# Check running containers
-docker-compose -f docker-compose.prod.yml ps
-
-# Restart a service
-docker-compose -f docker-compose.prod.yml restart backend
-
-# View real-time logs
-docker-compose -f docker-compose.prod.yml logs -f --tail=100
-
-# Execute commands in container
-docker-compose -f docker-compose.prod.yml exec backend sh
-
-# Database shell
-docker-compose -f docker-compose.prod.yml exec db psql -U picpeak
-
-# Clean restart
-docker-compose -f docker-compose.prod.yml down
-docker-compose -f docker-compose.prod.yml up -d
-```
-
-## Security Checklist
-
-- [ ] Strong JWT_SECRET (min 32 chars)
-- [ ] Strong database password
-- [ ] SSL/HTTPS enabled
-- [ ] Firewall configured (only 80/443 open)
-- [ ] Regular security updates
-- [ ] Backup encryption
-- [ ] Access logs monitored
-- [ ] Rate limiting enabled
-- [ ] File upload restrictions configured
-
-## Support
-
-For issues not covered here:
-1. Check application logs
-2. Review error messages carefully
-3. Ensure all environment variables are set
-4. Verify file permissions
-5. Check Docker daemon logs

README-LOCAL.md

@@ -1,130 +0,0 @@
-# 🚀 Quick Local Development Setup
-
-Get the photo sharing platform running locally in under 2 minutes!
-
-## Prerequisites
-- Docker Desktop installed and running
-- Git
-- 4GB RAM available
-
-## Quick Start
-
-```bash
-# 1. Clone the repository
-git clone <your-repo-url>
-cd picpeak
-
-# 2. Start everything
-./start-local.sh
-```
-
-That's it! 🎉
-
-## What You Get
-
-| Service | URL | Description |
-|---------|-----|-------------|
-| Frontend (Dev) | http://localhost:3002 | React app with hot reload |
-| Frontend (Prod) | http://localhost:3000 | Production build |
-| Backend API | http://localhost:3001 | Express API |
-| Mailhog | http://localhost:8025 | Email testing UI |
-
-## Default Credentials
-
-- **Admin Login**: Check `ADMIN_CREDENTIALS.txt` after first setup
-- **Test Gallery**: 
-  - Create via Admin Panel
-  - Set your own secure password
-
-## Common Tasks
-
-### View Logs
-```bash
-docker-compose -f docker-compose.local.yml logs -f
-```
-
-### Stop Everything
-```bash
-./stop-local.sh
-```
-
-### Reset Database
-```bash
-docker-compose -f docker-compose.local.yml exec backend npm run migrate
-```
-
-### Add Test Photos
-1. Create a gallery in the admin panel
-2. Get the gallery slug (e.g., `wedding-smith-2024`)
-3. Add photos to: `./storage/events/active/wedding-smith-2024/`
-4. Photos appear automatically!
-
-### Access Backend Shell
-```bash
-docker-compose -f docker-compose.local.yml exec backend sh
-```
-
-## Development Workflow
-
-1. **Frontend Development** (Port 3002)
-   - Hot reload enabled
-   - Edit files in `./frontend/src`
-   - Changes appear instantly
-
-2. **Backend Development** (Port 3001)
-   - Nodemon watches for changes
-   - Edit files in `./backend/src`
-   - Server restarts automatically
-
-3. **Email Testing**
-   - All emails go to Mailhog
-   - View at http://localhost:8025
-   - No real emails sent!
-
-## Troubleshooting
-
-### Backend won't start
-```bash
-# Check logs
-docker-compose -f docker-compose.local.yml logs backend
-
-# Rebuild
-docker-compose -f docker-compose.local.yml build backend
-```
-
-### Frontend build issues
-```bash
-# Clear cache and rebuild
-docker-compose -f docker-compose.local.yml exec frontend-dev npm run build
-```
-
-### Port conflicts
-Edit `docker-compose.local.yml` and change the port mappings:
-- Backend: Change `3001:3000` to `XXXX:3000`
-- Frontend: Change `3002:5173` to `YYYY:5173`
-
-### Reset everything
-```bash
-# Stop and remove all data
-docker-compose -f docker-compose.local.yml down -v
-rm -rf data storage logs
-./start-local.sh
-```
-
-## Tips
-
-- 📧 Check Mailhog for all emails
-- 🔄 Frontend auto-refreshes on save
-- 📁 SQLite DB at `./data/photo_sharing.db`
-- 🖼️ Photos in `./storage/events/active/`
-- 📝 Logs in `./logs/`
-
-## Next Steps
-
-1. Create your first gallery via Admin Panel
-2. Upload some test photos
-3. Test the gallery with password
-4. Check expiration warnings
-5. View emails in Mailhog
-
-Happy coding! 🎨

README.md

@@ -1,32 +1,576 @@
-# Photo Sharing Platform
+# 📸 PicPeak - Open Source Photo Sharing for Events
 
-A secure, self-hosted photo sharing platform designed for weddings and events. Features automatic expiration, email notifications, and simple file-based management.
+> [!IMPORTANT]
+> **PicPeak has moved to its own GitHub organization.**
+>
+> - **Docker images** are now published at `ghcr.io/picpeak/picpeak/{backend,frontend}`. The old path (`ghcr.io/the-luap/picpeak/...`) is no longer served — update your `docker-compose.yml`.
+> - **Branches**: active development is now on `main` (was `beta`); the curated stable channel is now `stable` (was `main`). Existing PRs and clones auto-redirect via GitHub.
+>
+> See **[`docs/migration-to-org.md`](docs/migration-to-org.md)** for the one-line `docker-compose.yml` edit and full details.
 
-## Features
+<div align="center">
+  <img src="docs/picpeak-logo.png" alt="PicPeak Logo" width="300" />
+  
+  [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
+  [![Docker](https://img.shields.io/badge/docker-%230db7ed.svg?style=flat&logo=docker&logoColor=white)](https://www.docker.com/)
+  [![Node.js](https://img.shields.io/badge/node.js-6DA55F?style=flat&logo=node.js&logoColor=white)](https://nodejs.org/)
+  [![React](https://img.shields.io/badge/react-%2320232a.svg?style=flat&logo=react&logoColor=%2361DAFB)](https://reactjs.org/)
+  [![Buy Me A Coffee](https://img.shields.io/badge/Buy%20Me%20a%20Coffee-theluap-FFDD00?logo=buymeacoffee&logoColor=black)](https://buymeacoffee.com/theluap)
 
-- 🔒 Password Protected Galleries
-- ⏰ Automatic Expiration
-- 📧 Email Notifications
-- 📁 Simple File Management
-- 📊 Analytics Integration
-- 🎨 Customizable Themes
-- 📱 Mobile Responsive
-- ⚡ Docker Ready
+  [Homepage](https://www.picpeak.app) · [Live Demo](https://demo.picpeak.app) · [Documentation](https://docs.picpeak.app) · [Support the project ☕](https://buymeacoffee.com/theluap)
+</div>
 
-## Quick Start
+**PicPeak** is a powerful, self-hosted open-source alternative to commercial photo-sharing platforms like PicDrop.com and Scrapbook.de. Designed specifically for photographers and event organizers, PicPeak makes it simple to share beautiful, time-limited photo galleries with clients while maintaining full control over your data and branding.
 
-1. Clone the repository
-2. Run `./scripts/install.sh`
-3. Configure `.env` file
-4. Setup SSL: `./scripts/setup-ssl.sh`
-5. Start: `docker-compose -f docker-compose.prod.yml up -d`
+![PicPeak Gallery Preview](docs/screenshot-gallery.png)
 
-Default credentials: Check ADMIN_CREDENTIALS.txt after first setup
+## 🎮 Live Demo
 
-## Documentation
+Try PicPeak without installing anything:
 
-See DEPLOYMENT.md for detailed deployment instructions.
+| | |
+|---|---|
+| **Demo URL** | [demo.picpeak.app](https://demo.picpeak.app) |
+| **Admin Panel** | [demo.picpeak.app/admin](https://demo.picpeak.app/admin) |
+| **Email** | `demo@picpeak.app` |
+| **Password** | `Demo2026!` |
 
-## License
+> The demo resets periodically. Uploaded content may be removed without notice.
 
-MIT License
+## 🌟 Why Choose PicPeak?
+
+Unlike expensive SaaS solutions, PicPeak gives you:
+
+- **💰 No Monthly Fees** - One-time setup, unlimited galleries
+- **🔒 Complete Data Control** - Your photos stay on your server
+- **🎨 White-Label Ready** - Full branding customization
+- **📱 Mobile-First Design** - Beautiful on all devices
+- **🚀 Lightning Fast** - Optimized performance and caching
+- **🌍 Multi-Language** - Built-in i18n support (EN, DE)
+
+## ✨ Key Features
+
+### For Photographers
+- 📁 **Drag & Drop Upload** - Simply drop photos into folders
+- 🔗 **External Media (Reference Mode)** - Browse and import from a read‑only external folder library without copying originals
+- ⏰ **Auto-Expiring Galleries** - Set expiration dates (default: 30 days)
+- 🔐 **Password Protection** - Secure client galleries
+- 📧 **Automated Emails** - Creation confirmations and expiration warnings
+- 📊 **Analytics Dashboard** - Track views, downloads, and engagement
+- 📽️ **Live Slideshow** - A separate fullscreen "Diashow" link per event for projectors at live events — auto-picks-up new uploads while it runs, with transitions, a logo watermark, and image-fit/colour options ([guide](docs/live-slideshow.md))
+- 🎨 **Custom Themes** - Match your brand perfectly
+- 🌐 **Public Landing Page** - Publish a curated marketing page when guests visit your root URL
+
+### For Clients
+- 🖼️ **Beautiful Galleries** - Clean, modern interface
+- 📱 **Mobile Optimized** - Swipe through photos on any device
+- ⬇️ **Bulk Downloads** - Download all photos with one click
+- 🔍 **Smart Search** - Find photos quickly
+- 📤 **Guest Uploads** - Optional client photo uploads
+- 🛡️ **Download Protection** - Advanced image protection with watermarking and right-click prevention
+
+### Technical Excellence
+- 🐳 **Docker Ready** - Deploy in minutes
+- 🔄 **Auto-Processing** - Automatic thumbnail generation
+- 🗂️ **Reference Library Support** - Point PicPeak at `EXTERNAL_MEDIA_ROOT` to reference existing originals, index quickly, and generate thumbnails on demand
+- 💾 **Smart Storage** - Automatic archiving of expired galleries
+- 🛡️ **Security First** - JWT auth, rate limiting, CORS protection
+- 📈 **Scalable** - From small studios to large agencies
+
+### For Studios — CRM & Accounting (Beta · off by default)
+- 📝 **Quotes → Contracts → Invoices** - One deal lineage; cancel-and-reissue (Storno) keeps issued invoices immutable
+- ⏱️ **Hours Logging & Calendar** - Per-customer time tracking; admin calendar of events, logged hours, and pending quotes/contracts
+- 🧾 **Inbound Supplier Invoices & Expenses** - Capture received invoices (upload/camera, rasterised server-side), categorise, and re-bill costs to clients
+- 📊 **Tax Report & Accountant Export** - Period-scoped income/cost report with VAT breakdown; PDF/CSV plus a Treuhänder/Banana (Swiss/LI) journal export, scopable to income-only or cost-only
+- 🌍 **VAT & Multi-currency** - Single VAT-code registry snapshotted onto each document; data-driven per-country rates
+- ⚠️ **Verify locally** - Feature-flagged off by default. Seeded contracts, QR/IBAN and tax defaults are **examples only** — review your own legal **and tax** regulations first (see disclaimers below)
+
+## 🚀 Quick Start
+
+Get PicPeak running in under 5 minutes:
+
+```bash
+# Clone the repository
+git clone https://github.com/PicPeak/picpeak.git
+cd picpeak
+
+# Copy the environment template — the defaults work out of the box.
+# Machine secrets (JWT, DB, Redis) are auto-generated on first run, and the
+# admin account is created in the browser (see below). Edit .env only to
+# customise (domain, SMTP, storage paths, …) — nothing is required.
+cp .env.example .env
+
+# Start with Docker Compose
+docker compose up -d
+
+# Access at http://localhost:3000
+```
+
+### First run — create your admin account
+
+On first start with no `ADMIN_PASSWORD` set, PicPeak has **no admin account yet** and greets you with an in-browser setup screen — no credentials in `.env`:
+
+1. Open **http://localhost:3000/admin** — you'll be redirected to `/setup`.
+2. Read the **one-time setup token** from the 0600 file the backend writes it to
+   (it is deliberately *not* printed to the logs — that would leave a live
+   bootstrap credential in `docker logs`):
+   ```bash
+   docker compose exec backend cat /app/data/SETUP_TOKEN
+   ```
+   It is bind-mounted, so `sudo cat data/SETUP_TOKEN` on the host works too. Only
+   if that file could not be written does the backend fall back to logging the
+   token (`docker compose logs backend | grep -i "setup token"`).
+3. Paste the token, set your admin **email + password**, and you're in. The token is single-use, and the setup screen closes permanently once an admin exists.
+
+> Prefer the old behaviour? Set `ADMIN_PASSWORD` in `.env` and PicPeak auto-creates the admin on first boot instead (credentials written to `data/ADMIN_CREDENTIALS.txt`).
+
+Note on Docker file permissions
+- The backend container starts as root, chowns bind-mounted host directories (`./storage`, `./data`, `./logs`) to UID 1001 (`nodejs`), then drops privileges via `su-exec` before running the app. No host-side setup needed for fresh installs.
+- If you pin `user:` in a compose override (e.g. to map a specific host UID), the self-chown is skipped and you must pre-chown the host directories to that UID — see [docs.picpeak.app/deployment/docker#permissions](https://docs.picpeak.app/deployment/docker#permissions).
+
+**ARM64 (aarch64) systems:** Pre-built images include native `linux/arm64`, no platform flags or emulation needed. If you're on an older image tag that's still amd64-only, see [docker-compose.amd64.override.yml](docker-compose.amd64.override.yml) for a transitional fallback.
+
+## 🔄 Release Channels
+
+PicPeak offers two release channels for different needs. Stable promotions are cut from a known-good beta point every 4–6 weeks — see [RELEASING.md](RELEASING.md) for the maintainer's promotion criteria and cadence policy.
+
+### Stable Channel (Recommended)
+- Production-ready releases
+- Thoroughly tested before release
+- Docker tags: `stable`, `latest`, or specific version like `v2.3.0`
+
+### Beta Channel
+- Early access to new features
+- May contain bugs or incomplete functionality
+- Docker tags: `beta` or specific version like `v2.3.0-beta.1`
+
+### Switching Channels
+
+Set the `PICPEAK_CHANNEL` environment variable in your `.env` file:
+
+```bash
+# For stable releases (default)
+PICPEAK_CHANNEL=stable
+
+# For beta releases
+PICPEAK_CHANNEL=beta
+
+# For a specific version
+PICPEAK_CHANNEL=v2.3.0
+```
+
+Then update your containers:
+
+```bash
+docker compose -f docker-compose.production.yml pull
+docker compose -f docker-compose.production.yml up -d
+```
+
+### Update Notifications
+
+The admin dashboard automatically notifies you when updates are available for your channel. To disable update checks, set:
+
+```bash
+UPDATE_CHECK_ENABLED=false
+```
+
+## 📖 Documentation
+
+Full documentation lives at **[docs.picpeak.app](https://docs.picpeak.app)** — deployment, admin settings reference, API docs, webhooks, archive lifecycle, branding, and everything else. Some quick links:
+
+- 🚀 [**Deployment**](https://docs.picpeak.app/deployment) - Docker, environment variables, reverse proxy, SSL
+- ⚙️ [**Admin Settings**](https://docs.picpeak.app/guides/admin-settings) - Every tab in the Settings panel
+- 🎯 [**Creating Events**](https://docs.picpeak.app/guides/creating-events) - Full event field reference
+- 📽️ [**Live Slideshow**](https://docs.picpeak.app/features/live-slideshow) - Fullscreen projector view that auto-updates during live events
+- 💾 [**Backup & Restore**](https://docs.picpeak.app/guides/backup-restore) - Backup configuration, restore wizard, full disaster recovery
+- 🔌 [**API Reference**](https://docs.picpeak.app/api) - REST endpoints, OpenAPI spec, webhooks
+- 🪝 [**Webhooks**](https://docs.picpeak.app/features/webhooks) - Event payloads, signing, filters, templates
+
+Project meta:
+
+- 🤝 [**Contributing**](CONTRIBUTING.md) - How to contribute
+- 📜 [**License**](LICENSE) - MIT License
+- 🔒 [**Security**](SECURITY.md) - Security policies
+- 📋 [**Code of Conduct**](CODE_OF_CONDUCT.md) - Community guidelines
+
+## 🌐 Public Landing Page
+
+Spotlight your studio with a customizable marketing page at `/`:
+
+- Head to **Admin → CMS Pages** to enable the public landing page toggle.
+- Edit the provided HTML template (rich sections, hero, testimonials) and optional CSS overrides.
+- The preview renders in a sandboxed iframe so you can iterate safely before publishing.
+- PicPeak sanitizes stored HTML and CSS server-side—scripts, iframes, and unsafe attributes are stripped automatically.
+- Use **Reset to default** anytime to restore the bundled template.
+- The backend caches the rendered landing page for 60 seconds by default; override with `PUBLIC_SITE_CACHE_TTL_MS` if you need a different TTL.
+- When the landing page is disabled PicPeak continues to serve the admin SPA/login exactly as before.
+
+## 🎯 Use Cases
+
+Perfect for:
+- 💒 **Wedding Photographers** - Share ceremony photos securely
+- 🎂 **Event Photography** - Birthday parties, corporate events
+- 📸 **Portrait Studios** - Client galleries with download limits
+- 🏢 **Corporate Events** - Internal photo sharing with branding
+- 🎓 **School Photography** - Secure parent access with expiration
+- 📽️ **Live Events** - Put a [Live Slideshow](docs/live-slideshow.md) on the venue projector that updates as you shoot
+
+## 🏗️ Tech Stack
+
+- **Backend**: Node.js, Express, SQLite/PostgreSQL
+- **Frontend**: React, Tailwind CSS, Framer Motion
+- **Storage**: Local filesystem (default) or S3-compatible object store (AWS S3, MinIO, R2, B2, Wasabi, Spaces) — see [Storage Backends](#storage-backends)
+- **Email**: SMTP with customizable templates
+- **Analytics**: Privacy-focused with Umami integration
+
+## 💾 Storage Backends
+
+PicPeak supports two storage backends for photos, thumbnails, hero images, watermarks, and archive zips. Both are configured via environment variables; no code change is required to switch.
+
+| Capability | `STORAGE_BACKEND=local` (default) | `STORAGE_BACKEND=s3` |
+|---|---|---|
+| Photo / thumbnail / hero storage | Local filesystem under `STORAGE_PATH` | Bucket on any S3-compatible service |
+| Admin UI upload | ✅ | ✅ |
+| Filesystem auto-import (chokidar watcher) | ✅ | ❌ — disabled (use the upload API) |
+| Watermarks, fingerprinting, fragmentation | ✅ | ✅ (materialized to a tmp file just-in-time) |
+| Bulk download zips (cached + on-the-fly) | ✅ | ✅ |
+| Backups | ✅ | ✅ |
+| External media reference mode (`EXTERNAL_MEDIA_ROOT`) | ✅ (always local) | ✅ (still local — not migrated) |
+
+### Switching to an S3-compatible backend
+
+1. Provision a bucket and credentials. The minimum IAM policy is documented in `.env.example`.
+2. Set `STORAGE_BACKEND=s3` plus `STORAGE_S3_BUCKET`, `STORAGE_S3_REGION`, `STORAGE_S3_ACCESS_KEY`, `STORAGE_S3_SECRET_KEY`. For non-AWS providers (MinIO, R2, B2, …) also set `STORAGE_S3_ENDPOINT`.
+3. If you have existing local content, copy it first: `node backend/scripts/migrate-storage.js --dry-run` then `node backend/scripts/migrate-storage.js`. The script is idempotent and writes a failures CSV.
+4. Restart the backend. The startup check pings the bucket and refuses to boot on misconfig.
+
+Note: presigned-URL serving (zero-bandwidth direct downloads from S3) is intentionally **not** in v1 — every request still streams through the backend so watermarks, devtools-detection, and access logging keep working.
+
+## 🔔 Webhooks
+
+PicPeak POSTs event/photo lifecycle notifications to URLs you configure under **Settings → Webhooks**. Each delivery is signed `HMAC-SHA256` with a per-webhook secret in the `X-PicPeak-Signature` header so receivers can verify the request really came from your PicPeak instance.
+
+### Event types
+
+| Event | Fires when |
+|---|---|
+| `event.created` | Gallery created (admin or API) |
+| `event.published` | Draft becomes live (`is_draft: true → false`) — also fires when an event is created with `is_draft=false` |
+| `event.archived` | Bulk-archive, manual archive, or auto-archive on expiry |
+| `event.expired` | Expiration checker marks the gallery inactive (fires before `event.archived` in the cascade) |
+| `photo.uploaded` | Admin upload, API upload, guest upload, or auto-import |
+| `photo.deleted` | Single delete, bulk delete (NOT fired per-photo when an event is archived — receivers infer from `event.archived` to avoid flooding) |
+
+### Payload shape
+
+```json
+{
+  "id": "delivery-uuid",
+  "type": "event.published",
+  "created_at": "2026-04-28T05:25:00.000Z",
+  "data": {
+    "event": { "id": 123, "slug": "wedding-smith", "share_url": "https://..." }
+  }
+}
+```
+
+Also sent on every request:
+- `X-PicPeak-Signature` — `HMAC-SHA256(secret, raw_body)` as hex
+- `X-PicPeak-Event` — the event type (handy for routing without parsing the body)
+- `X-PicPeak-Delivery` — UUID for idempotency on the receiver side
+- `User-Agent: PicPeak-Webhooks/1.0`
+
+### Verifying signatures
+
+**Node.js**
+```js
+const crypto = require('crypto');
+function verify(secret, rawBody, signature) {
+  const expected = crypto.createHmac('sha256', secret).update(rawBody).digest('hex');
+  const a = Buffer.from(expected, 'hex');
+  const b = Buffer.from(signature, 'hex');
+  if (a.length !== b.length) return false;
+  return crypto.timingSafeEqual(a, b);
+}
+```
+
+**Python**
+```python
+import hmac, hashlib
+def verify(secret: str, raw_body: bytes, signature: str) -> bool:
+    expected = hmac.new(secret.encode(), raw_body, hashlib.sha256).hexdigest()
+    return hmac.compare_digest(expected, signature)
+```
+
+**curl + openssl** (one-liner for a quick replay)
+```sh
+SIG=$(printf '%s' "$BODY" | openssl dgst -sha256 -hmac "$SECRET" | awk '{print $2}')
+[ "$SIG" = "$RECEIVED_SIG" ] && echo OK || echo MISMATCH
+```
+
+### Retries + observability
+
+- `2xx` → success, recorded with latency
+- Non-`2xx` or network error → exponential backoff: `1m → 5m → 30m → 2h → 12h`, max 5 attempts
+- After max attempts: status `failed`, surfaces in **Settings → Webhooks → Deliveries** with a "Replay" button
+- Up to 5 deliveries in flight at once; one slow consumer can't block others (configurable via `WEBHOOK_DELIVERY_CONCURRENCY`)
+- Response body truncated to 1KB before storage so chatty receivers don't bloat the audit log
+
+The deliveries page (`/admin/webhooks/:id/deliveries`) shows every attempt with timestamp, status, HTTP code, latency, payload sent, signature, and response. Click "Send test event" to fire a synthetic delivery for any event type.
+
+### SSRF protection
+
+Webhook URLs are validated against the same private-IP blocklist used elsewhere in the app — loopback, private RFC1918 ranges, link-local, `.local`/`.internal` hostnames, cloud metadata endpoints. The check runs both at create time and per-delivery (DNS-rebinding mitigation).
+
+For local development with a receiver on the same machine or docker network, set `WEBHOOK_ALLOW_PRIVATE_URLS=true`. Production deployments must leave this OFF.
+
+## 💻 System Requirements
+
+### Minimum Requirements
+- **CPU**: 2 CPU cores
+- **RAM**: **4 GB minimum** for a normal photo-upload workload — sharp/libvips
+  decodes the full uncompressed frame before resize, and the default two
+  worker loops at sharp-concurrency 2 can push peak RSS past 1.5 GB on a
+  batch of 20-MP+ photos. On a 2 GB VPS that's enough to OOM-kill the
+  backend mid-batch (surfaces as 503s on thumbnails — see [Low-memory
+  hosts](#low-memory-hosts) below for the recipe to run on 2 GB).
+- **Storage**: 20GB minimum (plus photo storage needs)
+- **OS**: Linux (Ubuntu 20.04+), macOS, or Windows with WSL2
+- **Node.js**: v18.0.0 or higher
+- **Database**: SQLite (included) or PostgreSQL 12+
+
+### Docker Requirements (Recommended)
+- **Docker**: v20.10.0+
+- **Docker Compose**: v2.0.0+
+
+### Low-memory hosts
+
+Running on 2 GB RAM (e.g. an entry-level VPS) is workable but requires
+tuning the upload-processor concurrency down. The backend auto-detects
+total RAM at startup via `os.totalmem()` — on a host that reports < 3 GB,
+it defaults `UPLOAD_PROCESSOR_CONCURRENCY` to **1** instead of 2 and logs
+a one-shot warning. You can pin the value explicitly in `.env`:
+
+```env
+# Single worker loop — slower batch processing, lower peak RSS
+UPLOAD_PROCESSOR_CONCURRENCY=1
+```
+
+The trade-off is throughput: a single worker processes one photo at a
+time, so a 100-photo batch takes ~2× as long but won't OOM. **Health-check
+note**: if the backend dies under memory pressure, the gallery serves
+`503 Service Unavailable` on thumbnails until Docker's
+`restart: unless-stopped` brings the container back. Persistent 503s
+during/after an upload batch on a low-memory host are almost always this.
+
+### Video Support Requirements
+When enabling video uploads, consider these additional resources:
+
+| Resource | Recommendation | Notes |
+|----------|----------------|-------|
+| **RAM** | 4GB+ recommended | FFmpeg processing requires more memory |
+| **Storage** | Plan for 10-100x more | Videos are significantly larger than images |
+| **CPU** | Additional cores help | Video thumbnail extraction is CPU-intensive |
+| **Bandwidth** | Higher throughput | Video streaming requires more bandwidth |
+
+**Technical Notes:**
+- FFmpeg is bundled via npm (`@ffmpeg-installer/ffmpeg`) - no system installation required
+- Maximum upload size: **10GB per video file**
+- Chunked upload support for files >100MB (resumable uploads)
+- Supported formats: MP4, WebM, MOV, AVI
+- Video thumbnails are automatically generated from the first few seconds
+
+**For Nginx/Reverse Proxy:**
+If using Nginx, increase the client max body size:
+```nginx
+client_max_body_size 10G;
+proxy_read_timeout 3600;
+proxy_send_timeout 3600;
+```
+
+## 🤝 Contributing
+
+We love contributions! PicPeak is built by photographers, for photographers. Whether you're fixing bugs, adding features, or improving documentation, your help is welcome.
+
+See our [Contributing Guide](CONTRIBUTING.md) for details.
+
+## 📊 Comparison with Alternatives
+
+| Feature | PicPeak | PicDrop | Scrapbook.de | Pixieset |
+|---------|---------|---------|--------------|----------|
+| Self-Hosted | ✅ | ❌ | ❌ | ❌ |
+| Custom Branding | ✅ Full | Limited | Limited | ✅ (paid) |
+| Monthly Cost | $0* | $29-199 | €19-99 | ~$60 |
+| Storage Limit | Unlimited** | 50-500GB | 100-1000GB | 3GB–Unlimited*** |
+| Client Uploads | ✅ | ✅ | ✅ | Limited |
+| API Access | ✅ | Paid | ❌ | ❌ |
+| Open Source | ✅ | ❌ | ❌ | ❌ |
+| Customer Accounts | ✅ | ❌ | ❌ | ✅ |
+| Quotes / Contracts / Invoices | 🧪 Beta | ❌ | ❌ | ✅ |
+| Incoming Invoices & Accounting | 🧪 Beta | ❌ | ❌ | ❌ |
+
+*You still bring your own server (own hardware or a VPS) and, if you want one, a domain.
+**Limited only by your server storage.
+***Pixieset's "unlimited" is photos only; video is capped by plan (roughly 0–10 h depending on tier).
+🧪 Beta = built but feature-flagged off by default (see [Beta Features](#-beta-features-use-at-your-own-risk)).
+
+## 🛡️ Security
+
+PicPeak takes security seriously:
+- 🔐 Password hashing with bcrypt
+- 🎫 JWT-based authentication
+- 🚦 Rate limiting on all endpoints
+- 🛡️ CORS protection
+- 📝 Activity logging
+- 🔒 Secure file access
+
+Found a security issue? Please open a [security issue](https://github.com/PicPeak/picpeak/issues/new?labels=security) on GitHub
+
+## 📸 Screenshots
+
+### 🎛️ **Admin Dashboard**
+Get a complete overview of your photo galleries, analytics, and system status.
+
+<img src="docs/screenshot-dashboard.png" alt="PicPeak Admin Dashboard" width="800" />
+
+### 📊 **Analytics & Insights**
+Track gallery performance, view statistics, and monitor user engagement.
+
+<img src="docs/screenshot-analytics.png" alt="PicPeak Analytics Dashboard" width="800" />
+
+### 📁 **Event Management**
+Organize and manage your photo galleries with intuitive event management tools.
+
+<img src="docs/screenshots-events.png" alt="PicPeak Events Management" width="800" />
+
+### ✨ **Key Interface Highlights**
+
+<details>
+<summary>👆 Click to see more interface details</summary>
+
+#### What makes PicPeak's interface special:
+
+- **🎨 Clean Design**: Modern, photographer-friendly interface
+- **📱 Responsive**: Perfect on desktop, tablet, and mobile
+- **⚡ Fast Loading**: Optimized for quick photo browsing
+- **🔒 Secure Access**: Password-protected galleries with expiration
+- **📤 Easy Uploads**: Drag & drop functionality for effortless photo management
+- **🎯 Client-Focused**: Intuitive gallery experience for your clients
+
+</details>
+
+## 🗺️ Roadmap
+
+We're constantly improving PicPeak and welcome contributions from our community! If you have ideas for new features or want to help implement existing ones, please open an issue or submit a pull request. Your contributions help make PicPeak better for everyone.
+
+### 🚧 Beta Features (Use at your own risk)
+
+These features are currently in beta testing and may have limited functionality or stability:
+
+| Feature | Description | Status |
+|---------|-------------|--------|
+| **CRM & Accounting Module** | Quotes, contracts, invoices (+ Storno), hours logging, calendar, and tax report — plus inbound supplier-invoice capture, internal expenses, and a Treuhänder/Banana (Swiss/LI) accountant-journal export. Feature-flagged off by default. Seeded contract blocks, payment terms, IBAN / QR-bill and tax defaults are **examples only** and need legal / financial / **tax** review before customer-facing use. See [docs.picpeak.app/features/crm](https://docs.picpeak.app/features/crm). | 🧪 Beta |
+| **Simple Deployment Script** | One-click deployment script for quick server setup with automated configuration and dependency installation | 🧪 Beta |
+
+### 📋 Future Enhancements
+
+| Feature | Description | Priority | Status |
+|---------|-------------|----------|---------|
+| **Backup & Restore** | Comprehensive backup system with S3/MinIO support, automated scheduling, and safe restore functionality | High | ✅ Implemented |
+| **External Media Library (Reference Mode)** | Use an external folder library as a read‑only source with import and on‑demand thumbnail generation | High | ✅ Implemented |
+| **Download Protection** | Advanced image protection system with canvas rendering, invisible watermarking, right-click prevention, and DevTools detection to protect photos from unauthorized downloads | High | ✅ Implemented |
+| **Gallery Templates** | Multiple gallery layouts (grid, masonry, carousel, timeline, hero, mosaic) with custom CSS styling support. Includes starter templates like Apple Liquid Glass for complete visual customization | Medium | ✅ Implemented |
+| **Face Recognition** | AI-powered face detection to help guests find their photos and create automatic person-based albums | Low | 🔄 Open |
+| **Gallery Feedback** | Allow guests to like, rate, and comment on photos with admin notifications and moderation | Medium | ✅ Implemented |
+| **Video Support** | Upload and display videos alongside photos in galleries with streaming support | Low | ✅ Implemented |
+| **Multiple Administrators** | Support for multiple admin accounts with role-based permissions and activity tracking | Low | ✅ Implemented |
+| **Filtering & Export Options** | Filter photos by likes, ratings, comments, or favorites. Search by filename. Sort by date, name, size, or rating. Export filtered selections as ZIP or generate Capture One/Lightroom-compatible file lists for professional workflows | Medium | ✅ Implemented |
+
+**Status Legend:** ✅ Implemented | 🚧 In Progress | 🔄 Open | 📋 Planned
+
+## ☕ Support the Project
+
+PicPeak is free, open source, and self-hostable forever. If it saves you time or replaces a paid subscription, consider buying me a coffee — it directly funds the time spent on new features, bug fixes, and keeping the demo + docs running.
+
+<p align="left">
+  <a href="https://buymeacoffee.com/theluap" target="_blank">
+    <img src="https://img.buymeacoffee.com/button-api/?text=Buy%20me%20a%20coffee&emoji=☕&slug=theluap&button_colour=FFDD00&font_colour=000000&font_family=Cookie&outline_colour=000000&coffee_colour=ffffff" alt="Buy Me A Coffee" />
+  </a>
+</p>
+
+Other ways to support without spending anything: ⭐ star the repo, share it with photographer friends, file good bug reports, or open a PR.
+
+## 🙏 Acknowledgments
+
+PicPeak is inspired by the best features of commercial platforms while remaining completely open source. Special thanks to all contributors who make this project possible.
+
+### 👥 Contributors
+
+A huge thank you to the people whose code, reports, and feedback have shaped PicPeak:
+
+- [**@the-luap**](https://github.com/the-luap) — creator and lead maintainer. Started the project and built PicPeak's foundation and the entire gallery experience (events, galleries, uploads, sharing, download protection, templates), plus backup & restore, analytics, system health, branding/theming, and WhatsApp notifications — and the architecture every later feature builds on.
+- [**@Luca-Timo**](https://github.com/Luca-Timo) — native Apple Silicon multi-arch images, external-URL toggle for legal CMS pages, the lazy-loaded folder tree picker, the admin-email picker on event creation, the data-driven self-hosted webfont system, the gallery header/banner decoupling, several typed-API refactors, and the CRM + accounting suite (quotes/contracts/invoices, hours logging, calendar, tax report, inbound supplier-invoice capture, expenses, and the Treuhänder/Banana export). Consistently raises the bar with thoughtful PRs.
+- [**@Rekoo-PS**](https://github.com/Rekoo-PS) — sharp-eyed bug reporter and product feedback. Filed the issues that drove the login-loop fix, the gallery-loading skeleton work, the redirection cleanup, the mobile-lightbox overhaul, the admin-events search-counter fix, the photo-count column, and the bulk-delete workflow. Also a [BuyMeACoffee](https://buymeacoffee.com/theluap) supporter — the kind of feedback loop that keeps the project useful for real deployments.
+
+If you've contributed and aren't listed here, please open a PR — this list is meant to grow.
+
+### 🤖 AI-Assisted Development
+
+This project was generated with the assistance of AI technology, but has been:
+- ✅ **Fully tested end-to-end** by human developers
+- 🔒 **Security audited** with comprehensive security checks
+- 👨‍💻 **Human-reviewed** for code quality and best practices
+- 🧪 **Production-tested** in real-world scenarios
+
+We believe in transparent development practices and the responsible use of AI as a tool to accelerate development while maintaining high standards of quality and security.
+
+## ⚠️ CRM & Accounting disclaimers — examples only, verify locally
+
+The CRM & accounting modules (contracts, invoices, QR-bills, the tax
+report and the accountant exports) ship seeded content and computed
+figures that are intended as a **starting point only**:
+
+- **Contract blocks** (image rights, NDA, model release, cancellation,
+  jurisdiction, …) are written by the maintainer, **not by a lawyer**.
+  Every operator must have their lawyer review and adapt them before
+  sending any contract to a customer.
+- **QR-bills and SEPA EPC payloads** are rendered from the data you
+  typed. Picpeak is open source — please scan a test invoice with your
+  bank's app to check the QR actually works. We are not responsible for
+  any mistakes that come from sending an invoice with bad data on it.
+- **Tax, VAT & accounting figures** (the tax report, VAT-payable, the
+  per-rate breakdown, the Treuhänder / Banana export, etc.) are computed
+  from the data you enter and the defaults you configure. They are
+  **guidance only and jurisdiction-specific** — tax rules, VAT rates,
+  deduction schemes (e.g. the Liechtenstein 20 % Gewinnungskosten flat
+  rate) and filing duties differ by country and change over time. **Every
+  operator must check their own tax / VAT regulations and verify the
+  numbers with their accountant / Treuhänder / tax authority before
+  relying on any figure or export.** Picpeak makes no warranty that the
+  output is correct for your jurisdiction or situation.
+
+Read [`docs/crm-disclaimers.md`](docs/crm-disclaimers.md) before
+enabling the Contracts, Invoices or Accounting features.
+
+## 📄 License
+
+PicPeak is released under the [MIT License](LICENSE). Use it freely for personal or commercial projects.
+
+## 🚀 Ready to Get Started?
+
+1. ⭐ **Star this repository** to show your support
+2. 📖 Read the [docs at docs.picpeak.app](https://docs.picpeak.app)
+3. 🐛 Report issues or request features
+4. 🤝 Join our community and contribute!
+
+---
+
+<p align="center">
+  Made with ❤️ by photographers, for photographers
+  <br>
+  <a href="https://www.picpeak.app">Homepage</a> •
+  <a href="https://demo.picpeak.app">Live Demo</a> •
+  <a href="https://github.com/PicPeak/picpeak">GitHub</a> •
+  <a href="https://docs.picpeak.app">Documentation</a> •
+  <a href="https://github.com/PicPeak/picpeak/issues">Support</a>
+</p>

RELEASING.md

@@ -0,0 +1,99 @@
+# Release Process
+
+This document describes how PicPeak releases are cut. It's the maintainer's reference, not user documentation — for the user-facing channel choice (stable vs pre-release) see the [Release Channels section in README.md](README.md#-release-channels).
+
+## TL;DR
+
+- **`main` branch** receives all merged work (active development). Every push triggers a `release-please` PR that proposes the next `vX.Y.Z-beta.N` pre-release. Merging that PR tags the pre-release and publishes Docker images under the `:main` rolling tag + the version-specific tag.
+- **`stable` branch** holds the curated stable channel. Stable releases are cut from a known-good `main` point via a `release/X.Y.Z-merge-from-main` branch and a manual PR to `stable`. Merging that PR triggers `release-please` to propose the stable release.
+- Target cadence: **a stable release every 4–6 weeks**, or sooner if `main` has been quiet and ready for promotion.
+
+> **Branch model background** — `main` (active dev) was previously called `beta`, and `stable` (curated channel) was previously called `main`. The rename happened with #669 to match the convention every other open-source project uses. The mechanics below all reference the post-rename names.
+
+## Cadence target
+
+4–6 weeks between stable releases is the working target. Reasoning:
+
+- Long enough that each stable carries meaningful changes worth the upgrade burden.
+- Short enough that pre-release users aren't carrying the "real" project alone for months — the stable channel should actually be usable as the recommended channel for new installs.
+- Aligns with how release-please surfaces pre-releases (multiple pre-release points usually accumulate inside a 4–6 week window, which gives natural promotion candidates).
+
+This is a target, not a hard rule. Cut sooner if `main` has been quiet and stable longer than usual. Cut later if `main` is in flux for security or migration reasons.
+
+## Promotion criteria
+
+A `main` tip is eligible for promotion to `stable` when **all** of the following hold:
+
+1. **CI green on the candidate `main` tip.** Specifically: `schema-drift` (`upgrade-from-bootstrap`), `fresh-install`, `Tests` (backend Jest + frontend Vitest), the four `Build and Push Docker Images` arch matrices, and `GitGuardian Security Checks`.
+2. **No open `bug`-labelled issues against the candidate for at least 7 days.** Issues fixed-but-not-yet-closed count as fixed; verify their PR is in the candidate `main` tip before closing them out.
+3. **An upgrade walk has been done on real production-shaped data** — apply the candidate's migration chain to a snapshot of the previous stable's DB and verify no manual intervention is required. CI proves fresh-install works; the upgrade walk is what proves the upgrade path works.
+4. **Operator-time smoke** on the candidate: log in, create event, upload photos, share gallery, open as a customer, log out. Catches binary-incompatibility regressions and UI-level breaks that unit tests don't see.
+
+If any of the four fail, the promotion waits. File any blockers as `bug`-labelled issues and let them bake on `main` before re-evaluating.
+
+## How a stable release is cut
+
+The actual mechanics, in order:
+
+1. **Pick the `main` tip.** Confirm it satisfies the four promotion criteria above. Note the exact SHA — that's what you're promoting.
+
+2. **Create the release branch from the `main` tip.**
+   ```bash
+   git push origin <main-tip-sha>:refs/heads/release/X.Y.Z-merge-from-main
+   ```
+   Naming convention: `release/X.Y.Z-merge-from-main`, where `X.Y.Z` is the stable version you intend to land. release-please will write the actual `X.Y.Z` on merge — the branch name is just a human label.
+
+3. **Open a PR to `stable`.** Title: `chore(release): promote main → stable as vX.Y.Z`. Body should summarise the major themes since the previous stable, the migration count, and any operator notes (e.g. "this release adds 22 migrations; existing installs should snapshot before upgrading"). See PR #568 as a worked example (predates the rename; the mechanics are unchanged).
+
+4. **Resolve conflicts.** `stable` almost always has commits `main` doesn't (security backports, release-please's stable-channel release commits, README rewrites). For each conflicting file, decide deliberately:
+   - **`backend/package.json` / `package-lock.json` + `frontend/package.json` / `package-lock.json`** — usually take `main`'s version (superset), but verify any security-pinned deps (`axios`, `nodemailer`, `i18next-http-backend`, `multer`, `tar`) on `main` are `>=` the pinned versions on `stable`. If `stable` has a newer pinned version (e.g. an emergency CVE backport `main` hasn't picked up), take `stable`'s pin.
+   - **`README.md`** — keep `stable`'s version if it has had a recent rewrite that `main` didn't pick up; otherwise take `main`'s.
+   - **`CHANGELOG.md`** — keep `stable`'s; release-please regenerates entries on its next stable cut from the commits going forward.
+   - **`.release-please-manifest.json`** — keep `stable`'s; release-please owns this file.
+   - Any other auto-merged file — spot-check that the auto-merge produced something sensible, especially for security-sensitive files (`backend/src/middleware/`, `backend/src/utils/tokenUtils.js`).
+
+5. **Wait for CI on the PR.** All ten checks (the original eight plus `merge-backend` and `merge-frontend`) must be green. If anything fails, fix on the release branch (NOT on `main` — `main` has already moved on).
+
+6. **Merge.** Standard merge commit, not squash — the PR's history (the individual feature commits) carries forward into `stable`'s log.
+
+7. **release-please picks it up.** Within minutes, release-please will open a new `chore(stable): release X.Y.Z` PR proposing the stable release. Review the auto-generated CHANGELOG.md entries for accuracy, edit if needed, and merge. That merge creates the `vX.Y.Z` git tag, publishes Docker images on the `:stable` and `:latest` tags, and creates the GitHub Release page.
+
+8. **Close the loop.** Bulk-close any `bug` issues that were fixed-but-not-closed and now appear in the released changelog. Reference the merge commit so reporters know which version contains the fix.
+
+## Hotfix path (backport to current stable)
+
+Regular bug fixes are generally backported automatically from `main` to `stable`. Keep backports focused on the fix, without unrelated features, and resolve conflicts manually when needed.
+
+**Security fixes are always released on both `stable` and `main`.** Do not wait for a full promotion to deliver a security update. A fix first applied to `stable` must also be forward-ported to `main`; a fix first applied to `main` must also reach `stable`. See [SECURITY.md](SECURITY.md) for the support policy.
+
+When a backport needs manual handling:
+
+1. Create a `security/cve-backport-X.Y.Z` or `fix/critical-X.Y.Z` branch off `stable`.
+2. Cherry-pick or hand-write the minimal fix.
+3. Open a PR to `stable` with the smallest possible diff.
+4. After merge, release-please will propose a patch-level stable release (e.g. `v3.55.1`).
+5. **Forward-port the fix to `main`** if it isn't already there. Otherwise the next full promotion will reintroduce the bug.
+6. For security fixes, verify that the fix has been published through **both** release channels; merging the code is only part of delivery.
+
+PR #412 ("backport 18 dependency CVE patches from beta") is a worked example of this path (predates the rename; the mechanics are unchanged).
+
+## Versioning
+
+PicPeak follows [Semantic Versioning](https://semver.org/) with one project-specific convention:
+
+- **MAJOR** bumps are reserved for breaking schema changes that require operator action on upgrade (e.g. a migration that's not safe to auto-apply, an env-var rename that can't be auto-detected).
+- **MINOR** bumps for new features, additive schema changes, and any change to the public HTTP API surface.
+- **PATCH** bumps for bug fixes and operator-invisible internal changes.
+- **Pre-release suffix** (`-beta.N`) for every `main`-channel cut; the `N` counter resets on each new MINOR or MAJOR target. The suffix kept the historical `-beta` literal even after the branch rename — operators were already pinning to `v3.x.y-beta.N` and changing the literal would have broken those pins.
+
+release-please derives all of this from conventional commit prefixes (`feat:`, `fix:`, `BREAKING CHANGE:`, etc.) automatically.
+
+## Things that don't go through this process
+
+- **Documentation-only changes** can land on either `stable` or `main` directly (no release cut needed); release-please will pick them up on the next regular release.
+- **Test-only changes** — same.
+- **CI / workflow changes** — same, but be aware they take effect on the branch they land on, so a CI fix targeting `main` won't fix a broken stable-channel workflow until the next promotion.
+
+## When this doc is wrong
+
+If you find yourself working around something here, update the doc before doing the workaround. The point of a written process is that future-you doesn't have to remember the workaround.

SECURITY.md

@@ -0,0 +1,81 @@
+# Security Policy
+
+## Scope
+
+This policy covers the PicPeak backend, frontend, all-in-one (AIO) image, optional
+ML component, and the Docker images published by the PicPeak project. Other
+PicPeak repositories define their own supported versions and release channels.
+
+## Supported Versions
+
+Security support follows the current release channels:
+
+| Version or channel | Security support |
+| --- | --- |
+| Latest stable release from `stable` | Supported; security fixes are published through this channel |
+| Latest beta release from `main` | Supported; security fixes are published through this channel |
+| Superseded stable or beta releases | Upgrade to the latest release in the same channel; older releases are not maintained separately |
+| 2.x and earlier | No longer supported |
+
+See the [latest stable release](https://github.com/PicPeak/picpeak/releases/latest)
+and [all releases, including betas](https://github.com/PicPeak/picpeak/releases).
+Version numbers differ between channels; each channel receives its own updates.
+
+### Security fixes and bug backports
+
+**Security fixes are always released on both `stable` and `main`.** A fix that
+lands on one branch must also reach the other branch and be published through
+both release channels. Security updates do not wait for the next full
+`main`-to-`stable` promotion.
+
+Regular bug fixes are also generally backported automatically to `stable`.
+Backports remain focused on the fix, without pulling in unrelated features.
+Maintainers resolve conflicts or handle a backport manually when necessary.
+
+The [release process](RELEASING.md) describes backports, forward-ports and
+publication. Operators must apply the published updates to their installations.
+
+## Reporting a Vulnerability
+
+**Do not report vulnerabilities in public issues, discussions or pull requests.**
+
+Report privately through:
+
+- [GitHub Private Vulnerability Reporting](https://github.com/PicPeak/picpeak/security/advisories/new) (preferred).
+- Email **info@picpeak.app** if you cannot use GitHub's private reporting form.
+
+Include the affected component, version or image tag, deployment method,
+reproduction steps, expected impact and any suggested fix. Share only the
+information needed to reproduce the problem; remove credentials and personal
+data from logs or examples.
+
+We aim to acknowledge reports within 48 hours. This is a response target, not a
+guaranteed service level or a promised resolution time. We will provide progress
+updates and coordinate disclosure with the reporter. Reporter credit is optional;
+tell us if you prefer to remain anonymous.
+
+## Deployment Security
+
+Security depends on both the software and its configuration. Operators should:
+
+- Use HTTPS and configure the reverse proxy and trusted proxy settings correctly.
+- Use strong credentials and keep deployment secrets private.
+- Apply updates for the chosen release channel and restrict unnecessary network access.
+- Keep backups and verify that they can be restored.
+
+See the deployment guides for [HTTPS](https://docs.picpeak.app/deployment/ssl-certificates),
+[reverse proxies](https://docs.picpeak.app/deployment/reverse-proxy),
+[security settings](https://docs.picpeak.app/guides/admin-settings/security)
+and [backup and restore](https://docs.picpeak.app/guides/backup-restore).
+
+## Vulnerability Disclosure
+
+We coordinate disclosure with the reporter while preparing fixes. Security fixes
+are published through both supported channels. Advisories and release notes
+identify affected versions, the fixed version in each channel, the impact and
+any required mitigation or upgrade steps. Reporter credit is included with
+permission.
+
+For ordinary bugs and support requests, use
+[GitHub Issues](https://github.com/PicPeak/picpeak/issues) or
+[GitHub Discussions](https://github.com/PicPeak/picpeak/discussions).

SETUP_GUIDE.md

@@ -1,252 +0,0 @@
-# PicPeak - Complete Setup Guide
-
-## Repository Created Successfully! 🎉
-
-Your PicPeak repository has been created at:
-**https://gitea.nothaft.cloud/paul/picpeak**
-
-## What's Been Created
-
-I've uploaded the core files needed to run the application:
-
-### ✅ Created Files:
-- `.gitignore` - Git ignore rules
-- `.dockerignore` - Docker ignore rules
-- `.env.example` - Environment configuration template
-- `docker-compose.yml` - Development Docker setup
-- `docker-compose.prod.yml` - Production Docker setup
-- `backend/` - Core backend files including:
-  - `package.json` - Dependencies
-  - `server.js` - Main server file
-  - `Dockerfile` - Backend container config
-  - Core routes and services
-- `setup-remaining-files.sh` - Script to create remaining files
-
-## Next Steps to Complete Setup
-
-### 1. Clone the Repository
-```bash
-git clone https://gitea.local.nothaft.cloud/paul/picpeak.git
-cd picpeak
-```
-
-### 2. Run the Setup Script
-```bash
-chmod +x setup-remaining-files.sh
-./setup-remaining-files.sh
-```
-
-This will create all remaining directories and files needed.
-
-### 3. Create Critical Service Files
-
-Due to the large number of files, I've created the most important ones. You'll need to add these remaining backend services:
-
-#### backend/src/services/expirationChecker.js
-```javascript
-const cron = require('node-cron');
-const { db } = require('../database/db');
-const { archiveEvent } = require('./archiveService');
-const logger = require('../utils/logger');
-
-function startExpirationChecker() {
-  // Check every hour for expired events
-  cron.schedule('0 * * * *', async () => {
-    await checkExpirations();
-  });
-  
-  logger.info('Expiration checker started');
-}
-
-async function checkExpirations() {
-  try {
-    const now = new Date();
-    const warningDate = new Date(now.getTime() + 7 * 24 * 60 * 60 * 1000);
-    
-    // Check for events needing warning emails
-    const eventsNeedingWarning = await db('events')
-      .where('is_active', true)
-      .where('is_archived', false)
-      .where('expires_at', '<=', warningDate)
-      .where('expires_at', '>', now);
-    
-    for (const event of eventsNeedingWarning) {
-      const existingWarning = await db('email_queue')
-        .where('event_id', event.id)
-        .where('email_type', 'warning')
-        .first();
-      
-      if (!existingWarning) {
-        await queueExpirationWarning(event);
-      }
-    }
-    
-    // Check for expired events
-    const expiredEvents = await db('events')
-      .where('is_active', true)
-      .where('is_archived', false)
-      .where('expires_at', '<=', now);
-    
-    for (const event of expiredEvents) {
-      await handleExpiredEvent(event);
-    }
-    
-  } catch (error) {
-    logger.error('Error checking expirations:', error);
-  }
-}
-
-async function queueExpirationWarning(event) {
-  const daysRemaining = Math.ceil((new Date(event.expires_at) - new Date()) / (1000 * 60 * 60 * 24));
-  
-  await db('email_queue').insert({
-    event_id: event.id,
-    recipient_email: event.host_email,
-    email_type: 'warning',
-    email_data: JSON.stringify({
-      event_name: event.event_name,
-      days_remaining: daysRemaining,
-      share_link: event.share_link
-    })
-  });
-  
-  logger.info(`Queued expiration warning for event ${event.slug}`);
-}
-
-async function handleExpiredEvent(event) {
-  try {
-    await db('events').where('id', event.id).update({ is_active: false });
-    
-    await db('email_queue').insert([
-      {
-        event_id: event.id,
-        recipient_email: event.host_email,
-        email_type: 'expiration',
-        email_data: JSON.stringify({
-          event_name: event.event_name
-        })
-      },
-      {
-        event_id: event.id,
-        recipient_email: event.admin_email,
-        email_type: 'expiration',
-        email_data: JSON.stringify({
-          event_name: event.event_name,
-          event_slug: event.slug
-        })
-      }
-    ]);
-    
-    await archiveEvent(event);
-    
-    logger.info(`Handled expiration for event ${event.slug}`);
-  } catch (error) {
-    logger.error(`Error handling expired event ${event.slug}:`, error);
-  }
-}
-
-module.exports = { startExpirationChecker };
-```
-
-### 4. Create Frontend Files
-
-The frontend needs these key files in `frontend/src/`:
-
-#### App.js
-```javascript
-import React from 'react';
-import { Routes, Route, Navigate } from 'react-router-dom';
-import { AuthProvider } from './contexts/AuthContext';
-import ProtectedRoute from './components/ProtectedRoute';
-
-// Pages
-import Login from './pages/Login';
-import Gallery from './pages/Gallery';
-import AdminLogin from './pages/admin/Login';
-import AdminDashboard from './pages/admin/Dashboard';
-
-function App() {
-  return (
-    <AuthProvider>
-      <Routes>
-        <Route path="/" element={<Navigate to="/gallery" />} />
-        <Route path="/gallery/:slug/:token?" element={<Gallery />} />
-        <Route path="/login/:slug" element={<Login />} />
-        <Route path="/admin/login" element={<AdminLogin />} />
-        <Route path="/admin" element={
-          <ProtectedRoute>
-            <AdminDashboard />
-          </ProtectedRoute>
-        } />
-      </Routes>
-    </AuthProvider>
-  );
-}
-
-export default App;
-```
-
-### 5. Install Dependencies
-
-```bash
-# Backend
-cd backend
-npm install
-
-# Frontend
-cd ../frontend
-npm install
-```
-
-### 6. Configure Environment
-
-Copy `.env.example` to `.env` and update with your settings:
-```bash
-cp .env.example .env
-nano .env
-```
-
-### 7. Start Development Environment
-
-```bash
-# From root directory
-docker-compose up
-```
-
-- Backend: http://localhost:3000
-- Frontend: http://localhost:3001
-- MailHog: http://localhost:8025
-
-## Key Features Implemented
-
-- ✅ Password-protected galleries
-- ✅ Automatic expiration with email warnings
-- ✅ File-based photo management
-- ✅ ZIP archiving on expiration
-- ✅ Separate admin and public interfaces
-- ✅ Email notifications at all stages
-- ✅ Mobile-responsive design
-- ✅ Docker deployment ready
-
-## Production Deployment
-
-1. Update `.env` with production values
-2. Run `./scripts/install.sh` on your server
-3. Configure SSL with `./scripts/setup-ssl.sh`
-4. Start with `docker-compose -f docker-compose.prod.yml up -d`
-
-## Need Help?
-
-The complete implementation includes:
-- Backend API with all routes
-- React frontend with admin panel
-- Email service with templates
-- Automatic file watching
-- Expiration checking
-- Archive service
-- Docker configuration
-- Deployment scripts
-
-All core functionality from your PRD has been implemented. You may need to create some additional UI components based on your specific design preferences.
-
-Default admin credentials: **admin / admin123** (change immediately!)

SIMPLE_SETUP.md

@@ -0,0 +1,568 @@
+# 🚀 PicPeak Simple Setup Guide
+
+This guide provides easy installation instructions for PicPeak on Linux servers with both Docker and non-Docker options.
+
+## 📋 Quick Start
+
+### One-Line Installation
+
+```bash
+# Download and run the unified setup script
+curl -fsSL https://raw.githubusercontent.com/PicPeak/picpeak/main/scripts/picpeak-setup.sh -o picpeak-setup.sh && \
+chmod +x picpeak-setup.sh && \
+sudo ./picpeak-setup.sh
+```
+
+The script will automatically detect your environment and recommend the best installation method.
+
+## 🎯 Installation Methods
+
+### Method 1: Docker Installation (Recommended)
+Best for: Most users, easy updates, isolated environment
+
+```bash
+sudo ./picpeak-setup.sh --docker
+```
+
+**Pros:**
+- ✅ Easier installation and updates
+- ✅ Better isolation from system
+- ✅ Consistent environment across platforms
+- ✅ Built-in PostgreSQL and Redis
+
+**Cons:**
+- ❌ Requires more resources (~4GB RAM recommended)
+- ❌ Additional Docker overhead
+
+### Method 2: Native Installation
+Best for: Resource-constrained systems, Raspberry Pi, direct control
+
+```bash
+sudo ./picpeak-setup.sh --native
+```
+
+**Pros:**
+- ✅ Lower resource usage (~1GB RAM minimum)
+- ✅ Direct system control
+- ✅ No Docker overhead
+- ✅ Better for ARM devices
+
+**Cons:**
+- ❌ More complex setup
+- ❌ System dependencies required
+- ❌ Manual update process
+
+## 📋 System Requirements
+
+### Minimum Requirements
+- **OS**: Ubuntu 20.04+, Debian 11+, Fedora 38+, RHEL/CentOS 8+, Raspberry Pi OS
+- **RAM**: 
+  - Docker: 2GB minimum (4GB recommended)
+  - Native: 1GB minimum (2GB recommended)
+- **Storage**: 2GB for application + space for photos
+- **Network**: Port 3001 (or 80/443 with proxy)
+
+### Supported Platforms
+- ✅ Ubuntu 20.04, 22.04, 24.04
+- ✅ Debian 11, 12
+- ✅ Raspberry Pi OS (32-bit and 64-bit)
+- ✅ Fedora 38, 39, 40
+- ✅ RHEL/CentOS/Rocky/AlmaLinux 8, 9
+
+## 🛠️ Installation Options
+
+### Interactive Mode (Default)
+```bash
+sudo ./picpeak-setup.sh
+```
+
+The script will prompt you to choose:
+1. Installation method (Docker or Native)
+2. Admin email and password
+3. Domain configuration (optional)
+4. Email server settings (optional)
+5. SSL/HTTPS setup (optional)
+
+### Unattended Installation
+
+#### Docker with full configuration:
+```bash
+sudo ./picpeak-setup.sh --docker --unattended \
+  --domain photos.example.com \
+  --email admin@example.com \
+  --admin-password SecurePass123 \
+  --smtp-host smtp.gmail.com \
+  --smtp-port 587 \
+  --smtp-user your-email@gmail.com \
+  --smtp-pass your-app-password \
+  --enable-ssl
+```
+
+#### Native with minimal configuration:
+```bash
+sudo ./picpeak-setup.sh --native --unattended \
+  --email admin@example.com \
+  --admin-password SecurePass123
+```
+
+### Command Line Options
+
+| Option | Description | Example |
+|--------|-------------|---------|
+| `--docker` | Use Docker installation | `--docker` |
+| `--native` | Use native installation | `--native` |
+| `--unattended` | Run without prompts | `--unattended` |
+| `--domain` | Domain for HTTPS setup | `--domain photos.example.com` |
+| `--email` | Admin email address | `--email admin@example.com` |
+| `--admin-password` | Set admin password | `--admin-password MySecurePass` |
+| `--smtp-host` | SMTP server hostname | `--smtp-host smtp.gmail.com` |
+| `--smtp-port` | SMTP server port | `--smtp-port 587` |
+| `--smtp-user` | SMTP username | `--smtp-user user@gmail.com` |
+| `--smtp-pass` | SMTP password | `--smtp-pass app-password` |
+| `--enable-ssl` | Enable HTTPS with Let's Encrypt | `--enable-ssl` |
+| `--port` | Custom port (native only) | `--port 8080` |
+| `--update` | Update existing installation | `--update` |
+| `--uninstall` | Remove installation | `--uninstall` |
+| `--help` | Show help message | `--help` |
+
+## 🏗️ What Gets Installed
+
+### Docker Installation
+```
+~/picpeak/                    # Or custom directory
+├── docker-compose.yml        # Service definitions
+├── .env                      # Configuration
+├── storage/
+│   └── events/              # Photo storage
+│       ├── active/          # Current galleries
+│       └── archived/        # Expired galleries
+├── logs/                    # Application logs
+└── backup/                  # Backup directory
+```
+
+**Services:**
+- PicPeak Backend (Node.js application)
+- PostgreSQL Database
+- Redis Cache
+- Nginx Reverse Proxy (optional)
+- Background Workers
+
+### Native Installation
+```
+/opt/picpeak/                # Installation directory
+├── backend/                 # Application code
+├── events/                  # Photo storage
+│   ├── active/             # Current galleries
+│   └── archived/           # Expired galleries
+├── logs/                   # Application logs
+└── config/                 # Configuration files
+```
+
+**Services (systemd):**
+- `picpeak-backend` - Main application
+- `picpeak-workers` - Background workers
+- `caddy` - Web server (optional)
+
+## 🔑 First Login — Create Your Admin
+
+If you installed with `picpeak-setup.sh` and gave an `--admin-password`, your admin account already exists — log in at `/admin` with that email and password.
+
+If you started PicPeak **without** setting `ADMIN_PASSWORD` (e.g. a plain `docker compose up`), there's **no admin yet** and you create it in the browser:
+
+1. Open `http://your-server:3000/admin` — you'll land on a setup screen.
+2. Read the **one-time setup token** from the 0600 file the backend writes it to
+   (it is not logged — that would leave a live credential in `docker logs`):
+   ```bash
+   docker compose exec backend cat /app/data/SETUP_TOKEN
+   ```
+   Only if that write fails does the backend log the token instead.
+3. Paste it, set your admin email + password. The token is single-use and the screen closes once an admin exists.
+
+## 🌐 Access Methods
+
+### Direct Access (Simplest)
+- Docker: `http://your-server:3000` (frontend and admin at `/admin`)
+- Backend/API: `http://your-server:3001` (API only; no UI routes)
+  
+For native installs, serve the built frontend (e.g., with nginx or Caddy) and access the admin at `/admin` on the frontend domain.
+
+### With Domain & HTTPS
+If configured during setup:
+- `https://your-domain.com` - Gallery frontend
+- `https://your-domain.com/admin` - Admin panel
+
+### Behind Existing Proxy
+Add to your Nginx/Apache configuration (split frontend vs backend):
+```nginx
+# Frontend (UI + /admin/*)
+location / {
+    proxy_pass http://localhost:3000;
+    proxy_http_version 1.1;
+    proxy_set_header Upgrade $http_upgrade;
+    proxy_set_header Connection 'upgrade';
+    proxy_set_header Host $host;
+    proxy_cache_bypass $http_upgrade;
+    proxy_set_header X-Real-IP $remote_addr;
+    proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
+    proxy_set_header X-Forwarded-Proto $scheme;
+}
+
+# Backend API and protected resources
+location /api {
+    proxy_pass http://localhost:3001;
+    proxy_http_version 1.1;
+    proxy_set_header Upgrade $http_upgrade;
+    proxy_set_header Connection 'upgrade';
+    proxy_set_header Host $host;
+    proxy_cache_bypass $http_upgrade;
+    proxy_set_header X-Real-IP $remote_addr;
+    proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
+    proxy_set_header X-Forwarded-Proto $scheme;
+    client_max_body_size 100M;
+}
+location ~ ^/(photos|thumbnails|uploads) {
+    proxy_pass http://localhost:3001;
+    proxy_http_version 1.1;
+    proxy_set_header Host $host;
+    proxy_set_header X-Real-IP $remote_addr;
+    proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
+    proxy_set_header X-Forwarded-Proto $scheme;
+}
+```
+
+## 📁 Managing Galleries
+
+### Creating a Gallery
+
+#### Via Admin Panel
+1. Login to admin panel at `/admin`
+2. Click "Create New Event"
+3. Configure settings (name, date, password, customer email)
+4. Upload photos via drag & drop in the Photos tab
+5. Publish the gallery when ready
+
+#### Adding Photos via File System
+
+> **Important:** You must first create the event in the admin panel. The file watcher only detects new photos for events that already exist in the database. You cannot create a gallery by copying files alone.
+
+Once an event exists, you can add photos by copying them into the event's folder. PicPeak's built-in file watcher will automatically detect the new files, create database records, and generate thumbnails.
+
+```bash
+# Docker installation — copy photos into an existing event's folder
+cp /path/to/photos/*.jpg ~/picpeak/storage/events/active/<event-slug>/
+
+# Native installation
+sudo cp /path/to/photos/*.jpg /opt/picpeak/events/active/<event-slug>/
+sudo chown -R picpeak:picpeak /opt/picpeak/events/active/<event-slug>
+```
+
+The event slug is visible in the admin panel URL or share link (e.g. `wedding-smith-2024`). Supported formats: `.jpg`, `.jpeg`, `.png`, `.webp`. The file watcher has a 2-second stability delay before processing new files.
+
+### Gallery Structure
+```
+<event-slug>/
+├── collages/        # Group photos (optional subfolder)
+├── individual/      # Individual photos (optional subfolder)
+└── photo.jpg        # Photos at root level also work
+```
+
+## 🔧 Service Management
+
+### Docker Installation
+
+```bash
+cd ~/picpeak
+
+# Check status
+docker compose ps
+
+# View logs
+docker compose logs -f
+
+# Stop services
+docker compose down
+
+# Start services
+docker compose up -d
+
+# Restart services
+docker compose restart
+
+# Update PicPeak
+docker compose pull
+docker compose up -d
+```
+
+### Native Installation
+
+```bash
+# Check status
+sudo systemctl status picpeak-backend
+sudo systemctl status picpeak-workers
+
+# View logs
+sudo journalctl -u picpeak-backend -f
+sudo journalctl -u picpeak-workers -f
+
+# Start services
+sudo systemctl start picpeak-backend picpeak-workers
+
+# Stop services
+sudo systemctl stop picpeak-backend picpeak-workers
+
+# Restart services
+sudo systemctl restart picpeak-backend picpeak-workers
+
+# Update PicPeak
+# (reruns migrations to pick up schema fixes for native installs)
+sudo ./picpeak-setup.sh --update
+```
+
+## ⚙️ Configuration
+
+### Docker Configuration
+Edit `~/picpeak/.env`:
+```bash
+nano ~/picpeak/.env
+docker compose restart
+```
+
+### Native Configuration
+Edit `/opt/picpeak/app/backend/.env`:
+```bash
+sudo nano /opt/picpeak/app/backend/.env
+sudo systemctl restart picpeak-backend
+```
+
+### Key Settings
+
+| Setting | Description | Default |
+|---------|-------------|---------|
+| `JWT_SECRET` | Token signing secret | Auto-generated |
+| `ADMIN_EMAIL` | Admin email | admin@example.com |
+| `ADMIN_PASSWORD` | Admin password | Auto-generated |
+| `PHOTOS_DIR` | Photo storage path | Varies by method |
+| `SMTP_ENABLED` | Email notifications | false |
+| `DEFAULT_EXPIRY_DAYS` | Gallery expiration | 30 |
+
+## 📧 Email Configuration
+
+### Gmail Setup
+1. Enable 2-Factor Authentication
+2. Generate App Password
+3. Configure:
+```env
+SMTP_ENABLED=true
+SMTP_HOST=smtp.gmail.com
+SMTP_PORT=587
+SMTP_SECURE=false
+SMTP_USER=your-email@gmail.com
+SMTP_PASS=your-app-password
+SMTP_FROM=noreply@yourdomain.com
+```
+
+### SendGrid Setup
+1. Sign up at sendgrid.com (100 emails/day free)
+2. Create API key
+3. Configure:
+```env
+SMTP_ENABLED=true
+SMTP_HOST=smtp.sendgrid.net
+SMTP_PORT=587
+SMTP_USER=apikey
+SMTP_PASS=your-sendgrid-api-key
+SMTP_FROM=verified-sender@yourdomain.com
+```
+
+## 🔄 Maintenance
+
+### Backups
+
+#### Docker:
+```bash
+# Backup script included
+cd ~/picpeak
+./backup.sh
+
+# Manual backup
+docker exec picpeak-postgres pg_dump -U picpeak picpeak > backup.sql
+tar -czf photos-backup.tar.gz storage/events/
+```
+
+#### Native:
+```bash
+# Database backup
+sudo cp /opt/picpeak/app/backend/data/photo_sharing.db /backup/database-$(date +%Y%m%d).sqlite
+
+# Photos backup
+sudo tar -czf /backup/photos-$(date +%Y%m%d).tar.gz /opt/picpeak/events/
+```
+
+### Updates
+
+```bash
+# Docker
+cd ~/picpeak
+docker compose pull
+docker compose up -d
+
+# Native
+sudo ./picpeak-setup.sh --update
+```
+
+### Uninstall
+
+```bash
+# Will prompt for confirmation and data removal options
+sudo ./picpeak-setup.sh --uninstall
+```
+
+## 🐛 Troubleshooting
+
+### Common Issues
+
+#### Service Won't Start
+```bash
+# Docker
+docker compose logs backend
+docker compose down && docker compose up -d
+
+# Native
+sudo journalctl -u picpeak-backend -n 50
+sudo systemctl restart picpeak-backend
+```
+
+#### Can't Access Admin Panel
+1. Check firewall:
+```bash
+# Ubuntu/Debian
+sudo ufw allow 3001
+
+# RHEL/CentOS
+sudo firewall-cmd --add-port=3001/tcp --permanent
+sudo firewall-cmd --reload
+```
+
+2. Verify service:
+```bash
+# Docker
+curl http://localhost:3001/api/health
+
+# Native
+sudo systemctl is-active picpeak-backend
+```
+
+#### Photos Not Showing
+```bash
+# Check permissions (Native)
+sudo chown -R picpeak:picpeak /opt/picpeak/events/
+sudo chmod -R 755 /opt/picpeak/events/
+
+# Check permissions (Docker)
+ls -la ~/picpeak/storage/events/
+```
+
+#### Reset Admin Password
+
+```bash
+# Docker
+docker exec picpeak-backend node scripts/reset-admin-password.js
+
+# Native
+cd /opt/picpeak/app/backend
+sudo -u picpeak node scripts/reset-admin-password.js
+```
+
+> **Note:** The new password will be displayed in the console output and saved to `ADMIN_PASSWORD_RESET.txt`. Save it immediately!
+
+### Getting Help
+
+1. **Check logs:**
+   - Docker: `docker compose logs -f`
+   - Native: `sudo journalctl -u picpeak-backend -f`
+   - Installation: `/tmp/picpeak-setup-*.log`
+
+2. **Documentation:**
+   - [Full Documentation](https://docs.picpeak.app)
+   - [Deployment Guide](https://docs.picpeak.app/deployment)
+
+3. **Support:**
+   - [GitHub Issues](https://github.com/PicPeak/picpeak/issues)
+   - Include: Error messages, system info (`uname -a`), installation method
+
+## 🔒 Security Best Practices
+
+### Essential Security
+1. **Change default admin password immediately**
+2. **Use HTTPS for production** (Let's Encrypt included)
+3. **Configure firewall** (only open necessary ports)
+4. **Regular updates** (system and PicPeak)
+5. **Automated backups** (configure in admin panel)
+
+### Advanced Security
+- Use VPN for admin panel access
+- Configure fail2ban for brute force protection
+- Enable audit logging
+- Regular security scans
+- Implement IP whitelisting
+
+## 📊 Performance Optimization
+
+### Docker Optimization
+```yaml
+# Adjust in docker-compose.yml
+services:
+  backend:
+    deploy:
+      resources:
+        limits:
+          cpus: '2'
+          memory: 2G
+```
+
+### Native Optimization
+```bash
+# Increase Node.js memory
+echo "NODE_OPTIONS=--max-old-space-size=2048" >> /opt/picpeak/app/backend/.env
+sudo systemctl restart picpeak-backend
+```
+
+## 🎯 Quick Setup Examples
+
+### Home/Office Network
+```bash
+# Simple local setup without domain
+sudo ./picpeak-setup.sh --native --email admin@local.com
+```
+
+### Public Website with HTTPS
+```bash
+# Full production setup
+sudo ./picpeak-setup.sh --docker \
+  --domain photos.company.com \
+  --email admin@company.com \
+  --enable-ssl
+```
+
+### Raspberry Pi Setup
+```bash
+# Optimized for ARM devices
+sudo ./picpeak-setup.sh --native \
+  --port 8080 \
+  --email pi@local.com
+```
+
+## ✅ Post-Installation Checklist
+
+- [ ] Admin password changed
+- [ ] Email configuration tested
+- [ ] First test gallery created
+- [ ] Backup schedule configured
+- [ ] Firewall rules applied
+- [ ] SSL certificate working (if applicable)
+- [ ] Monitoring setup
+- [ ] Documentation bookmarked
+
+---
+
+**PicPeak Setup v1.0** | [Documentation](https://github.com/PicPeak/picpeak) | [Support](https://github.com/PicPeak/picpeak/issues)

TODO.md

@@ -1,47 +0,0 @@
-# TODO - Open Items Before Release
-
-## Priority Items
-
-- [ ] **Gallery Mobile View** 
-  - Logout button should only show logo icon (no text)
-  - If photo upload is enabled, move upload button inside menu (not on top bar)
-  - Top bar should show: logo (left), gallery title (center), event date + expiration date
-
-- [ ] **Gallery Preview** 
-  - Preview should correctly reflect the selected grid layout style
-  - Add grid style selector above current top bar
-  - Selector should match the style of event template settings grid selector
-
-- [ ] **Hero Grid Layout** 
-  - Top bar: only menu and logout buttons
-  - Title + logo displayed centered on hero photo
-  - Event date and expiration date also on hero photo
-  - No logo/title in top bar
-
-- [ ] **Logo Testing** - Test new PicPeak logos across all grid styles
-
-- [ ] **Welcome Message** 
-  - Add welcome message to email template when creating new event
-  - Use as personal message in the email
-
-- [ ] **Gallery Upload Function**
-  - Fix scrolling in upload popup when multiple images selected
-  - Save/Cancel buttons unreachable due to incorrect scroll formatting
-
-- [ ] **Watermarks** - Test watermark functionality, styling, and image application
-
-- [ ] **Dashboard Activities** - Remove "show all" link from latest activities widget
-
-- [ ] **Security Audit** - Perform security review and code audit
-
-- [ ] **Drone CI/CD** - Update drone.yaml configuration
-
-- [ ] **Version Management** - Implement automatic version updates on commits/builds
-
-## Completed Items
-
-_(Move completed items here with date)_
-
----
-
-Last updated: 2025-07-10

TRAEFIK_DEPLOYMENT.md

@@ -1,280 +0,0 @@
-# Traefik Deployment Guide
-
-This guide explains how to deploy the PicPeak application with Traefik as the reverse proxy.
-
-## Overview
-
-The application consists of:
-- **Frontend**: React app served by nginx (port 80)
-- **Backend**: Node.js API (port 3000)
-- **Database**: PostgreSQL (port 5432, internal only)
-
-## Traefik Configuration
-
-### 1. Docker Labels for Traefik
-
-Add these labels to your `docker-compose.prod.yml` services:
-
-```yaml
-services:
-  frontend:
-    labels:
-      - "traefik.enable=true"
-      - "traefik.http.routers.picpeak-frontend.rule=Host(`picpeak.yourdomain.com`)"
-      - "traefik.http.routers.picpeak-frontend.entrypoints=websecure"
-      - "traefik.http.routers.picpeak-frontend.tls.certresolver=letsencrypt"
-      - "traefik.http.services.picpeak-frontend.loadbalancer.server.port=80"
-      # Priority for catch-all route
-      - "traefik.http.routers.picpeak-frontend.priority=1"
-
-  backend:
-    labels:
-      - "traefik.enable=true"
-      - "traefik.http.routers.picpeak-api.rule=Host(`picpeak.yourdomain.com`) && PathPrefix(`/api`)"
-      - "traefik.http.routers.picpeak-api.entrypoints=websecure"
-      - "traefik.http.routers.picpeak-api.tls.certresolver=letsencrypt"
-      - "traefik.http.services.picpeak-api.loadbalancer.server.port=3000"
-      # Higher priority for API routes
-      - "traefik.http.routers.picpeak-api.priority=10"
-      
-      # Additional routes for backend static files
-      - "traefik.http.routers.picpeak-uploads.rule=Host(`picpeak.yourdomain.com`) && PathPrefix(`/uploads`)"
-      - "traefik.http.routers.picpeak-uploads.entrypoints=websecure"
-      - "traefik.http.routers.picpeak-uploads.tls.certresolver=letsencrypt"
-      - "traefik.http.routers.picpeak-uploads.service=picpeak-api"
-      - "traefik.http.routers.picpeak-uploads.priority=10"
-      
-      - "traefik.http.routers.picpeak-images.rule=Host(`picpeak.yourdomain.com`) && PathPrefix(`/images`)"
-      - "traefik.http.routers.picpeak-images.entrypoints=websecure"
-      - "traefik.http.routers.picpeak-images.tls.certresolver=letsencrypt"
-      - "traefik.http.routers.picpeak-images.service=picpeak-api"
-      - "traefik.http.routers.picpeak-images.priority=10"
-```
-
-### 2. Network Configuration
-
-Ensure your services are on the Traefik network:
-
-```yaml
-networks:
-  picpeak:
-    external: false
-  traefik:
-    external: true
-
-services:
-  frontend:
-    networks:
-      - picpeak
-      - traefik
-      
-  backend:
-    networks:
-      - picpeak
-      - traefik
-      
-  db:
-    networks:
-      - picpeak  # Don't expose to traefik
-```
-
-### 3. Remove Nginx Service
-
-Since you're using Traefik, remove the nginx service from `docker-compose.prod.yml`:
-
-```yaml
-# Remove this entire service:
-# nginx:
-#   image: nginx:alpine
-#   ...
-```
-
-## Frontend Configuration
-
-The frontend is built with the API URL set to `/api`. This is important because:
-
-1. All API calls will be relative to the same domain
-2. Traefik will route `/api/*` to the backend service
-3. No CORS issues since everything is on the same domain
-
-## Environment Variables
-
-Ensure these are set correctly:
-
-```bash
-# Backend needs to know the public URLs
-ADMIN_URL=https://picpeak.yourdomain.com
-FRONTEND_URL=https://picpeak.yourdomain.com
-
-# Backend API is accessed via /api path
-API_URL=https://picpeak.yourdomain.com/api
-```
-
-## Complete Example
-
-Here's a complete `docker-compose.prod.yml` for Traefik:
-
-```yaml
-version: '3.8'
-
-networks:
-  picpeak:
-    external: false
-  traefik:
-    external: true
-
-services:
-  backend:
-    image: picpeak-backend:latest
-    build:
-      context: ./backend
-      dockerfile: Dockerfile
-    restart: unless-stopped
-    depends_on:
-      - db
-    environment:
-      - NODE_ENV=production
-      - PORT=3000
-      - JWT_SECRET=${JWT_SECRET}
-      - ADMIN_URL=https://picpeak.yourdomain.com
-      - FRONTEND_URL=https://picpeak.yourdomain.com
-      - DATABASE_CLIENT=pg
-      - DB_HOST=db
-      - DB_PORT=5432
-      - DB_USER=${DB_USER:-picpeak}
-      - DB_PASSWORD=${DB_PASSWORD}
-      - DB_NAME=${DB_NAME:-picpeak}
-      - SMTP_HOST=${SMTP_HOST}
-      - SMTP_PORT=${SMTP_PORT}
-      - SMTP_SECURE=${SMTP_SECURE}
-      - SMTP_USER=${SMTP_USER}
-      - SMTP_PASS=${SMTP_PASS}
-      - EMAIL_FROM=${EMAIL_FROM}
-      - STORAGE_PATH=/app/storage
-      - EVENTS_PATH=/app/storage/events
-      - ARCHIVE_PATH=/app/storage/events/archived
-    volumes:
-      - ./storage:/app/storage
-      - ./data:/app/data
-      - ./logs:/app/logs
-    networks:
-      - picpeak
-      - traefik
-    labels:
-      - "traefik.enable=true"
-      - "traefik.http.routers.picpeak-api.rule=Host(`picpeak.yourdomain.com`) && PathPrefix(`/api`)"
-      - "traefik.http.routers.picpeak-api.entrypoints=websecure"
-      - "traefik.http.routers.picpeak-api.tls.certresolver=letsencrypt"
-      - "traefik.http.services.picpeak-api.loadbalancer.server.port=3000"
-      - "traefik.http.routers.picpeak-api.priority=10"
-      
-      - "traefik.http.routers.picpeak-uploads.rule=Host(`picpeak.yourdomain.com`) && PathPrefix(`/uploads`)"
-      - "traefik.http.routers.picpeak-uploads.entrypoints=websecure"
-      - "traefik.http.routers.picpeak-uploads.tls.certresolver=letsencrypt"
-      - "traefik.http.routers.picpeak-uploads.service=picpeak-api"
-      - "traefik.http.routers.picpeak-uploads.priority=10"
-      
-      - "traefik.http.routers.picpeak-images.rule=Host(`picpeak.yourdomain.com`) && PathPrefix(`/images`)"
-      - "traefik.http.routers.picpeak-images.entrypoints=websecure"
-      - "traefik.http.routers.picpeak-images.tls.certresolver=letsencrypt"
-      - "traefik.http.routers.picpeak-images.service=picpeak-api"
-      - "traefik.http.routers.picpeak-images.priority=10"
-
-  frontend:
-    image: picpeak-frontend:latest
-    build:
-      context: ./frontend
-      dockerfile: Dockerfile
-      args:
-        - VITE_API_URL=/api
-    restart: unless-stopped
-    depends_on:
-      - backend
-    networks:
-      - picpeak
-      - traefik
-    labels:
-      - "traefik.enable=true"
-      - "traefik.http.routers.picpeak-frontend.rule=Host(`picpeak.yourdomain.com`)"
-      - "traefik.http.routers.picpeak-frontend.entrypoints=websecure"
-      - "traefik.http.routers.picpeak-frontend.tls.certresolver=letsencrypt"
-      - "traefik.http.services.picpeak-frontend.loadbalancer.server.port=80"
-      - "traefik.http.routers.picpeak-frontend.priority=1"
-
-  db:
-    image: postgres:14-alpine
-    restart: unless-stopped
-    environment:
-      - POSTGRES_USER=${DB_USER:-picpeak}
-      - POSTGRES_PASSWORD=${DB_PASSWORD}
-      - POSTGRES_DB=${DB_NAME:-picpeak}
-      - POSTGRES_HOST_AUTH_METHOD=scram-sha-256
-      - POSTGRES_INITDB_ARGS=--auth-host=scram-sha-256 --auth-local=trust
-    volumes:
-      - postgres_data:/var/lib/postgresql/data
-    networks:
-      - picpeak
-    command: postgres -c ssl=off
-
-volumes:
-  postgres_data:
-```
-
-## Troubleshooting
-
-### 502 Bad Gateway Errors
-
-1. **Check if backend is running**:
-   ```bash
-   docker-compose -f docker-compose.prod.yml ps
-   docker-compose -f docker-compose.prod.yml logs backend
-   ```
-
-2. **Verify Traefik can reach the backend**:
-   - Ensure both services are on the same Docker network
-   - Check Traefik logs: `docker logs traefik`
-
-3. **Check backend health**:
-   ```bash
-   docker-compose -f docker-compose.prod.yml exec backend curl http://localhost:3000/api/health
-   ```
-
-### Frontend Can't Reach API
-
-1. **Verify API paths don't have double `/api`**:
-   - Frontend should call `/auth/admin/login`, not `/api/auth/admin/login`
-   - The base URL in axios should be `/api`
-
-2. **Check browser console for actual URLs being called**
-
-3. **Ensure Traefik routing rules are correct**:
-   - API routes should have higher priority than frontend catch-all
-
-### CORS Issues
-
-Should not occur since everything is on the same domain. If you see CORS errors:
-1. Check that `FRONTEND_URL` and `ADMIN_URL` match your actual domain
-2. Ensure you're not mixing HTTP and HTTPS
-
-## Testing the Setup
-
-1. **Test API directly**:
-   ```bash
-   curl https://picpeak.yourdomain.com/api/health
-   ```
-
-2. **Test frontend**:
-   ```bash
-   curl https://picpeak.yourdomain.com/
-   ```
-
-3. **Test admin login**:
-   - Navigate to https://picpeak.yourdomain.com/admin/login
-   - Check browser console for any errors
-
-## Important Notes
-
-1. **SSL/TLS**: Traefik handles SSL termination, so the backend doesn't need SSL
-2. **Port Exposure**: Don't expose backend ports directly - let Traefik handle routing
-3. **Health Checks**: Configure Traefik health checks for better reliability
-4. **Rate Limiting**: Consider adding Traefik rate limiting middleware for API routes

TROUBLESHOOTING-TRAEFIK.md

@@ -1,134 +0,0 @@
-# Traefik Troubleshooting Guide
-
-## Common Issues and Solutions
-
-### 1. 404 Errors on API Routes
-
-**Problem**: Getting 404 errors when accessing `/api/*` routes
-
-**Causes**:
-- Traefik routing rules not properly configured
-- Backend container not healthy
-- Path stripping not working correctly
-
-**Solutions**:
-
-1. **Check container health**:
-```bash
-docker ps # Check if backend is running
-docker logs picpeak-backend # Check for startup errors
-```
-
-2. **Test backend directly**:
-```bash
-# Access backend container
-docker exec -it picpeak-backend sh
-
-# Test health endpoint
-wget -O- http://localhost:3000/health
-
-# Test public settings endpoint
-wget -O- http://localhost:3000/public/settings
-```
-
-3. **Check Traefik routing**:
-```bash
-# Check if routes are registered in Traefik
-curl https://traefik.yourdomain.com/api/http/routers | jq '.[] | select(.rule | contains("picpeak"))'
-```
-
-### 2. Backend Not Accessible Through Traefik
-
-**Key Configuration Points**:
-
-1. **Traefik Labels** (in deploy section):
-   - `traefik.enable=true` - Enable Traefik for this container
-   - `traefik.docker.network=proxy` - Specify which network Traefik should use
-   - `traefik.http.routers.picpeak-backend.priority=100` - Higher priority for API routes
-
-2. **Path Stripping**:
-   - Frontend expects `/api/*` but backend serves routes without `/api` prefix
-   - Middleware strips `/api` before forwarding to backend
-
-3. **Network Configuration**:
-   - Backend must be in both `picpeak` (internal) and `proxy` (Traefik) networks
-
-### 3. Environment Variable Issues
-
-**Critical Variables**:
-- `ADMIN_URL` and `FRONTEND_URL` must match your actual domain
-- These affect CORS configuration
-
-**Example .env**:
-```env
-# URLs
-ADMIN_URL=https://picpeak.local.nothaft.cloud
-FRONTEND_URL=https://picpeak.local.nothaft.cloud
-
-# Database
-DB_USER=picpeak
-DB_PASSWORD=your_secure_password
-DB_NAME=picpeak
-
-# JWT
-JWT_SECRET=your_secure_jwt_secret
-
-# Email (optional)
-SMTP_HOST=smtp.example.com
-SMTP_PORT=587
-SMTP_SECURE=true
-SMTP_USER=noreply@example.com
-SMTP_PASS=smtp_password
-EMAIL_FROM=noreply@example.com
-```
-
-### 4. Debugging Steps
-
-1. **Check if backend is receiving requests**:
-```bash
-# Watch backend logs
-docker logs -f picpeak-backend
-
-# Look for incoming requests when you try to access the admin page
-```
-
-2. **Test API routes directly**:
-```bash
-# From outside
-curl -v https://picpeak.local.nothaft.cloud/api/public/settings
-
-# Should see backend logs if request reaches container
-```
-
-3. **Verify Traefik middleware**:
-```bash
-# Check if stripprefix middleware exists
-curl https://traefik.yourdomain.com/api/http/middlewares | jq '.[] | select(.name | contains("picpeak"))'
-```
-
-### 5. Quick Fix Checklist
-
-- [ ] Backend container is healthy (`docker ps`)
-- [ ] Backend is in both networks (`docker inspect picpeak-backend | grep -A 20 Networks`)
-- [ ] Traefik labels use correct network (`traefik.docker.network=proxy`)
-- [ ] Priority is set correctly (backend: 100, frontend: 10)
-- [ ] ADMIN_URL and FRONTEND_URL match your domain
-- [ ] Database is accessible from backend
-- [ ] Migrations have run successfully
-
-### 6. Alternative Testing
-
-If Traefik routing is problematic, test backend directly:
-
-```bash
-# Port forward to test backend directly
-docker run --rm -it --network picpeak alpine/curl curl http://backend:3000/health
-
-# Or expose backend port temporarily
-docker run -d --name picpeak-backend-test \
-  --network picpeak \
-  -p 3001:3000 \
-  registry.local.nothaft.cloud/picpeak-backend:latest
-```
-
-Then access http://localhost:3001/health to verify backend is working.

backend/.env.example

@@ -3,39 +3,114 @@
 
 # Application
 NODE_ENV=production
-PORT=3000
+PORT=3001
 
 # Security
-JWT_SECRET=your-very-secure-jwt-secret-at-least-32-characters-long
+# Generate with: openssl rand -base64 32
+JWT_SECRET=your-very-secure-jwt-secret-at-least-32-characters-long-example123456
 
-# URLs
-ADMIN_URL=https://yourdomain.com
-FRONTEND_URL=https://yourdomain.com
+# Admin 2FA (TOTP) secret encryption key — OPTIONAL.
+# Admin authenticator secrets are encrypted at rest (AES-256-GCM). By default
+# the key is derived from JWT_SECRET, so you do NOT need to set this. Set it
+# only if you want the MFA encryption key decoupled from JWT_SECRET (e.g. so
+# rotating JWT_SECRET doesn't invalidate enrolled authenticators). If you set
+# it, changing/losing it makes existing 2FA secrets undecryptable — recover
+# with: docker compose exec backend node scripts/reset-admin-mfa.js --all --yes
+# Generate with: openssl rand -base64 32
+#MFA_ENCRYPTION_KEY=
+
+# Auth cookie Secure flag
+#   unset  - default: 'auto' in production, false in dev (#427)
+#   true   - always set Secure (HTTPS-only cookies; breaks plain-HTTP access —
+#            login appears to succeed but the browser silently drops the
+#            cookie, leaving you in a redirect loop. Only set this if you
+#            ALWAYS reach the site via HTTPS)
+#   false  - never set Secure (allows HTTP; cookies not protected on HTTPS)
+#   auto   - decide per request: Secure on HTTPS, not on HTTP. Reads
+#            req.secure from Express which respects X-Forwarded-Proto from a
+#            trusted reverse proxy. This is the default and is the right
+#            choice for most deployments.
+#
+# Why 'auto' is the default in production:
+#   - On real HTTPS (reverse proxy with X-Forwarded-Proto), req.secure is
+#     true → Secure flag is still emitted. No security regression vs. true.
+#   - On plain HTTP (LAN access, first-time install before reverse proxy is
+#     wired up), req.secure is false → Secure flag is omitted → login works
+#     instead of silently looping back to /admin/login.
+#
+# When you'd set this explicitly:
+#   - COOKIE_SECURE=true → strict HTTPS-only deployments where you want
+#     defense in depth against accidentally serving over HTTP.
+#   - COOKIE_SECURE=false → you intentionally only ever serve over HTTP and
+#     don't want the per-request check (rare).
+#
+# Requirements for 'auto' mode to detect HTTPS correctly:
+#   1. Your reverse proxy MUST send X-Forwarded-Proto: https on HTTPS
+#      requests. Standard configs for NPM/Traefik/Caddy do this by default.
+#   2. The proxy must be on a trusted IP range. By default PicPeak trusts
+#      loopback and private networks (127.0.0.1, 10.x, 172.16-31.x,
+#      192.168.x, link-local). Proxies outside those ranges need custom
+#      trust proxy configuration.
+# COOKIE_SECURE=auto
+
+# Cookie SameSite attribute (Lax | Strict | None). Default: Lax
+# COOKIE_SAMESITE=Lax
+
+# Cookie Domain — set this if serving auth cookies across subdomains.
+# Leave unset for same-origin setups.
+# COOKIE_DOMAIN=.example.com
+
+# URLs (adjust for your domain)
+ADMIN_URL=https://photos.example.com
+FRONTEND_URL=https://photos.example.com
+BACKEND_URL=https://photos.example.com  # Or https://api.photos.example.com if separate
+
+# API URL for email assets (logos, images in emails)
+# This must be the publicly accessible URL where recipients can load images
+# If not set, defaults to http://localhost:3001 which will break images in production emails
+API_URL=https://photos.example.com/api
 
 # Database Configuration
 DATABASE_CLIENT=pg
-DB_HOST=db
+DB_HOST=localhost
 DB_PORT=5432
 DB_USER=picpeak
-DB_PASSWORD=your-secure-database-password
+DB_PASSWORD=your-secure-database-password-change-this
 DB_NAME=picpeak
 
-# Email Configuration
-SMTP_HOST=smtp.example.com
+# Email Configuration (Examples for common providers)
+# Gmail example:
+# SMTP_HOST=smtp.gmail.com
+# SMTP_PORT=587
+# SMTP_SECURE=false
+# SMTP_USER=your-email@gmail.com
+# SMTP_PASS=your-app-specific-password
+
+# SendGrid example:
+SMTP_HOST=smtp.sendgrid.net
 SMTP_PORT=587
 SMTP_SECURE=false
-SMTP_USER=your-smtp-username
-SMTP_PASS=your-smtp-password
-EMAIL_FROM=noreply@yourdomain.com
+SMTP_USER=apikey
+SMTP_PASS=your-sendgrid-api-key
+EMAIL_FROM=noreply@example.com
 
-# Storage Paths (Docker)
+# Storage Paths
+# IMPORTANT: STORAGE_PATH must be set to avoid file path resolution issues
+# Docker deployment:
 STORAGE_PATH=/app/storage
 EVENTS_PATH=/app/storage/events
 ARCHIVE_PATH=/app/storage/events/archived
 
-# Analytics (Optional)
-UMAMI_URL=https://analytics.yourdomain.com
-UMAMI_WEBSITE_ID=your-website-id
+# Local development:
+# STORAGE_PATH=./storage
+# EVENTS_PATH=./storage/events
+# ARCHIVE_PATH=./storage/events/archived
+
+# Analytics Backend Configuration (OPTIONAL)
+# Used for server-side tracking only
+# Primary configuration should be done through Admin UI > Settings > Analytics
+# UMAMI_URL=https://analytics.example.com
+# UMAMI_WEBSITE_ID=b4d3c2a1-5678-90ab-cdef-1234567890ab
 
 # Logging
 LOG_LEVEL=info

backend/Dockerfile

@@ -1,23 +1,86 @@
-FROM node:18-alpine AS builder
+FROM node:22-alpine AS builder
+
+# Add build arguments
+ARG CACHEBUST=1
+ARG BUILD_DATE
+ARG VCS_REF
+ARG VERSION
+
+# Add labels for GitHub Container Registry
+LABEL org.opencontainers.image.source="https://github.com/PicPeak/picpeak"
+LABEL org.opencontainers.image.description="PicPeak Backend Service"
+LABEL org.opencontainers.image.licenses="MIT"
 
 WORKDIR /app
 
 # Copy package files
 COPY package*.json ./
 
-# Install dependencies
-RUN npm ci --only=production
+# Install dependencies (--omit=dev replaces deprecated --only=production)
+RUN npm ci --omit=dev
 
 # Copy application files
 COPY . .
 
 # Production stage
-FROM node:18-alpine
+FROM node:22-alpine
 
 WORKDIR /app
 
-# Install dumb-init for proper signal handling and postgresql-client for database checks
-RUN apk add --no-cache dumb-init postgresql-client
+# knexfile.js picks its config block by NODE_ENV, and the `development` block
+# defaults to sqlite3. Leaving NODE_ENV unset here meant every deployment that
+# doesn't go through our compose files — Kubernetes, Helm, plain `docker run` —
+# silently ran on SQLite and ignored DB_HOST/DB_USER/DB_PASSWORD, while
+# wait-for-db.sh (shell, reads DB_HOST directly) reported "PostgreSQL is up" in
+# the same log. The compose files still override this, so nothing changes for
+# compose users. See #1038.
+ENV NODE_ENV=production
+
+# Redeclare CACHEBUST — ARGs don't cross stage boundaries, so the builder
+# stage's declaration never reached this stage. Consuming it in the RUN below
+# busts that layer's cache every CI run (CACHEBUST=github.run_number), so the
+# image always picks up current Alpine security updates instead of reusing a
+# stale cached upgrade layer.
+ARG CACHEBUST=1
+
+# Upgrade all packages to fix security vulnerabilities (OpenSSL, libexpat, BusyBox CVEs)
+RUN echo "cachebust=${CACHEBUST}" && apk upgrade --no-cache
+
+# Remove the npm CLI from the final image. Nothing runs npm here: the
+# entrypoint is node, runtime deps are COPY'd from the builder stage, and
+# wait-for-db.sh invokes the migration runners via node directly. npm's
+# bundled node_modules kept tripping Trivy (sigstore, tar 7.5.19,
+# brace-expansion 5.0.7 — even npm 12.0.1 still ships the vulnerable
+# copies), so shipping no npm ends that alert class instead of chasing
+# per-release patches. Note: `docker exec … npm run <script>` no longer
+# works in the container — use `node migrations/run-migrations-safe.js`
+# and friends instead.
+RUN rm -rf /usr/local/lib/node_modules/npm /usr/local/bin/npm /usr/local/bin/npx
+
+# Install dumb-init for proper signal handling, postgresql-client for database
+# checks, ffmpeg for video upload support, and su-exec for the root → nodejs
+# privilege drop in wait-for-db.sh (see #484: container starts as root so it
+# can chown bind-mounted host volumes to UID 1001, then re-execs as nodejs
+# before running the app). Alpine's ffmpeg package ships both `ffmpeg` and
+# `ffprobe` built natively against musl libc — the npm
+# `@ffmpeg-installer/ffmpeg` binary is glibc-built and (a) doesn't reliably
+# run on Alpine and (b) only includes ffmpeg, not ffprobe (which the video
+# pipeline calls via fluent-ffmpeg.ffprobe()).
+# fontconfig is required so `sharp` (librsvg) can rasterise SVG logos that
+# contain live <text> for the CRM PDFs. Without any font installed, librsvg
+# renders text as tofu boxes (□) while the vector artwork still draws — i.e.
+# a "corrupted" logo on invoices/quotes. DejaVu/Liberation provide a broad
+# Unicode fallback; picpeak's own brand fonts (assets/fonts/, the same files
+# PDFKit + the web UI use) are registered with fontconfig further down so the
+# logo's text renders in its actual typeface, not a fallback.
+# poppler-utils provides `pdftoppm`, used to rasterise inbound supplier-invoice
+# PDFs to flat PNGs server-side so the admin UI NEVER renders a raw (possibly
+# malicious) PDF. pdftoppm does not execute embedded JS or fetch remote
+# resources, so it doubles as the SSRF/phone-home guard for untrusted inbound
+# documents (see docs/accounting-inbound-invoices.md).
+RUN apk add --no-cache dumb-init postgresql-client ffmpeg su-exec \
+    fontconfig ttf-dejavu ttf-liberation poppler-utils && \
+    fc-cache -f
 
 # Create non-root user
 RUN addgroup -g 1001 -S nodejs && adduser -S nodejs -u 1001
@@ -26,16 +89,33 @@ RUN addgroup -g 1001 -S nodejs && adduser -S nodejs -u 1001
 COPY --from=builder --chown=nodejs:nodejs /app/node_modules ./node_modules
 COPY --chown=nodejs:nodejs . .
 
-# Make wait script executable
-RUN chmod +x wait-for-db.sh
+# Ensure all source files are readable and wait script is executable
+RUN chmod -R a+r /app && chmod +x wait-for-db.sh
+
+# Register picpeak's bundled brand fonts (assets/fonts/<Family>/*.ttf — the
+# same files PDFKit and the web UI use) with fontconfig, so when sharp/librsvg
+# rasterises an SVG logo its <text> renders in the actual brand typeface
+# rather than a DejaVu/Liberation fallback. fontconfig indexes by each font's
+# internal family name and recurses into the per-family subdirectories.
+RUN printf '<?xml version="1.0"?>\n<!DOCTYPE fontconfig SYSTEM "fonts.dtd">\n<fontconfig>\n  <dir>/app/assets/fonts</dir>\n</fontconfig>\n' > /etc/fonts/conf.d/99-picpeak-fonts.conf && \
+    fc-cache -f /app/assets/fonts
 
 # Create necessary directories
 RUN mkdir -p storage/events/active storage/events/archived storage/thumbnails data logs && \
     chown -R nodejs:nodejs storage data logs
 
-USER nodejs
+# No USER directive — the container starts as root so wait-for-db.sh can
+# chown bind-mounted host directories to UID 1001 before dropping privs
+# via su-exec. See #484 for the fresh-install restart loop this avoids.
 
 EXPOSE 3000
 
+# Healthcheck hits the same /health endpoint already used by the e2e
+# runner and by the docker-compose `depends_on: condition: service_healthy`
+# checks. wget is part of the Alpine base image. Long start-period covers
+# the wait-for-db.sh delay before the Node process starts listening.
+HEALTHCHECK --interval=30s --timeout=5s --start-period=60s --retries=3 \
+  CMD wget --no-verbose --tries=1 --spider http://localhost:3000/health || exit 1
+
 ENTRYPOINT ["dumb-init", "--"]
 CMD ["./wait-for-db.sh", "node", "server.js"]

backend/Dockerfile.dev

@@ -1,9 +1,14 @@
-FROM node:18-alpine
+FROM node:20-alpine
 
 WORKDIR /app
 
-# Install dumb-init for proper signal handling
-RUN apk add --no-cache dumb-init
+# Upgrade all packages to fix security vulnerabilities (BusyBox CVEs)
+RUN apk upgrade --no-cache
+
+# Install dumb-init for proper signal handling and ffmpeg for video uploads.
+# Alpine's ffmpeg ships both ffmpeg + ffprobe built natively against musl;
+# the npm-bundled binary doesn't run reliably on Alpine. Match production.
+RUN apk add --no-cache dumb-init ffmpeg
 
 # Copy package files
 COPY package*.json ./
@@ -25,5 +30,8 @@ USER nodejs
 
 EXPOSE 3000
 
+HEALTHCHECK --interval=30s --timeout=5s --start-period=30s --retries=3 \
+  CMD wget --no-verbose --tries=1 --spider http://localhost:3000/health || exit 1
+
 ENTRYPOINT ["dumb-init", "--"]
 CMD ["npm", "run", "dev"]

backend/__tests__/README.md

@@ -0,0 +1,229 @@
+# Enhanced Backup System Test Suite
+
+This directory contains comprehensive tests for the enhanced backup system with S3 support.
+
+## Test Structure
+
+### Unit Tests
+- `services/backupService.enhanced.test.js` - Unit tests for the enhanced backup service
+  - Configuration management
+  - S3 backup functionality
+  - Manifest generation
+  - Error handling and recovery
+  - Backward compatibility (local and rsync)
+  - Service lifecycle management
+
+### Integration Tests
+- `integration/backup-s3.test.js` - Integration tests for S3 backups
+  - Real S3/MinIO connection tests
+  - Full backup process with actual files
+  - Incremental backup verification
+  - Manifest storage and retrieval
+  - Error recovery scenarios
+
+### Manual Integration Test Script
+- `../scripts/test-backup-integration.js` - Comprehensive manual testing script
+  - Can test against MinIO, AWS S3, or any S3-compatible service
+  - Tests all backup types (S3, local, rsync)
+  - Performance testing with large files
+  - Detailed progress reporting
+
+## Running Tests
+
+### Prerequisites
+
+1. **For Unit Tests**: No special setup required, all dependencies are mocked.
+
+2. **For Integration Tests**: Requires a running S3-compatible service (MinIO recommended)
+   ```bash
+   # Start MinIO using Docker
+   docker run -d \
+     -p 9000:9000 \
+     -p 9001:9001 \
+     --name minio-test \
+     -e MINIO_ROOT_USER=minioadmin \
+     -e MINIO_ROOT_PASSWORD=minioadmin \
+     minio/minio server /data --console-address ":9001"
+   ```
+
+3. **Environment Variables** (for integration tests):
+   ```bash
+   # Optional - defaults work with local MinIO
+   export TEST_S3_ENDPOINT=http://localhost:9000
+   export TEST_S3_ACCESS_KEY=minioadmin
+   export TEST_S3_SECRET_KEY=minioadmin
+   
+   # Skip S3 tests if no S3 service available
+   export SKIP_S3_TESTS=true
+   ```
+
+### Running Unit Tests
+
+```bash
+# Run all backup service tests
+npm test -- __tests__/services/backupService.enhanced.test.js
+
+# Run specific test suite
+npm test -- __tests__/services/backupService.enhanced.test.js -t "S3 Backup Functionality"
+
+# Run with coverage
+npm test -- --coverage __tests__/services/backupService.enhanced.test.js
+```
+
+### Running Integration Tests
+
+```bash
+# Ensure MinIO is running first!
+
+# Run S3 integration tests
+npm test -- __tests__/integration/backup-s3.test.js
+
+# Run with verbose output
+npm test -- __tests__/integration/backup-s3.test.js --verbose
+
+# Skip S3 tests if needed
+SKIP_S3_TESTS=true npm test -- __tests__/integration/backup-s3.test.js
+```
+
+### Running Manual Integration Tests
+
+```bash
+# Test with local MinIO (default)
+node scripts/test-backup-integration.js
+
+# Test with AWS S3
+node scripts/test-backup-integration.js \
+  --endpoint https://s3.amazonaws.com \
+  --access-key YOUR_ACCESS_KEY \
+  --secret-key YOUR_SECRET_KEY \
+  --bucket your-test-bucket
+
+# Test local backup
+node scripts/test-backup-integration.js --type local
+
+# Test with cleanup after completion
+node scripts/test-backup-integration.js --cleanup
+
+# Verbose output
+node scripts/test-backup-integration.js --verbose
+```
+
+## Test Coverage
+
+The test suite covers:
+
+### Configuration
+- ✅ Database configuration retrieval
+- ✅ JSON parsing and error handling
+- ✅ Configuration validation
+- ✅ Required field validation
+
+### S3 Functionality
+- ✅ S3 client initialization
+- ✅ Connection testing
+- ✅ File upload with progress tracking
+- ✅ Large file handling (multipart upload)
+- ✅ Metadata and custom headers
+- ✅ Error handling and retries
+
+### Backup Process
+- ✅ Full backup execution
+- ✅ Incremental backup (changed files only)
+- ✅ File checksum calculation and comparison
+- ✅ Database backup inclusion
+- ✅ Archive inclusion toggle
+- ✅ File size limits
+
+### Manifest Generation
+- ✅ Full manifest generation
+- ✅ Incremental manifest with parent reference
+- ✅ JSON and YAML format support
+- ✅ Manifest validation
+- ✅ S3 manifest storage and retrieval
+- ✅ Checksum verification
+
+### Error Handling
+- ✅ S3 connection failures
+- ✅ File read errors
+- ✅ Individual file failure recovery
+- ✅ Retry logic with exponential backoff
+- ✅ Email notifications on failure
+- ✅ Concurrent backup prevention
+
+### Backward Compatibility
+- ✅ Local directory backup
+- ✅ Rsync backup
+- ✅ Existing manifest format support
+
+### Service Management
+- ✅ Cron job scheduling
+- ✅ Service start/stop
+- ✅ Manual backup triggering
+- ✅ Backup history and status
+
+## Mock Setup
+
+The unit tests use comprehensive mocking:
+
+```javascript
+// Database mocking
+jest.mock('../../src/database/db');
+
+// S3 client mocking
+jest.mock('../../src/services/storage/s3Storage');
+
+// File system mocking
+const mockFs = require('mock-fs');
+
+// Cron job mocking
+jest.mock('node-cron');
+```
+
+## CI/CD Integration
+
+To run tests in CI/CD pipeline:
+
+```yaml
+# Example GitHub Actions
+- name: Run Unit Tests
+  run: npm test -- __tests__/services/backupService.enhanced.test.js
+
+- name: Start MinIO
+  run: |
+    docker run -d \
+      -p 9000:9000 \
+      --name minio-test \
+      -e MINIO_ROOT_USER=minioadmin \
+      -e MINIO_ROOT_PASSWORD=minioadmin \
+      minio/minio server /data
+
+- name: Run Integration Tests
+  run: npm test -- __tests__/integration/backup-s3.test.js
+```
+
+## Debugging Tests
+
+```bash
+# Run tests in debug mode
+node --inspect-brk ./node_modules/.bin/jest __tests__/services/backupService.enhanced.test.js
+
+# Run single test with console output
+npm test -- __tests__/services/backupService.enhanced.test.js -t "should perform S3 backup" --verbose
+```
+
+## Performance Considerations
+
+- Integration tests create real files and S3 objects
+- Each test run creates a unique S3 bucket to avoid conflicts
+- Cleanup is automatic but can be disabled for debugging
+- Large file tests (10MB+) are included but can be slow
+
+## Adding New Tests
+
+When adding new backup features:
+
+1. Add unit tests to `backupService.enhanced.test.js`
+2. Add integration tests to `backup-s3.test.js` if S3-specific
+3. Update manual test script for comprehensive testing
+4. Ensure mocks are properly configured
+5. Document any new environment requirements

backend/__tests__/adminSettings.logo.test.js

@@ -0,0 +1,184 @@
+const fs = require('fs');
+const fsPromises = fs.promises;
+const os = require('os');
+const path = require('path');
+const express = require('express');
+const request = require('supertest');
+
+describe('Admin settings logo upload flow', () => {
+  let tmpDir;
+  let router;
+  let app;
+  let settingsStore;
+
+  const resetModules = () => {
+    jest.resetModules();
+    jest.clearAllMocks();
+  };
+
+  beforeEach(async () => {
+    resetModules();
+
+    tmpDir = await fsPromises.mkdtemp(path.join(os.tmpdir(), 'picpeak-logo-'));
+    process.env.STORAGE_PATH = tmpDir;
+
+    settingsStore = new Map();
+
+    const buildQuery = (table) => {
+      const filters = [];
+      const applyFilters = (rows) => {
+        if (filters.length === 0) {
+          return rows;
+        }
+        return rows.filter((row) =>
+          filters.every(({ column, value }) => row[column] === value)
+        );
+      };
+
+      const makeRow = (row) => ({ ...row });
+
+      return {
+        where(column, value) {
+          filters.push({ column, value });
+          return this;
+        },
+        first() {
+          if (table === 'app_settings') {
+            const rows = applyFilters(Array.from(settingsStore.values()).map(makeRow));
+            return Promise.resolve(rows[0]);
+          }
+          return Promise.resolve(undefined);
+        },
+        select() {
+          return Promise.resolve([]);
+        },
+        sum() {
+          return Promise.resolve({ total: 0 });
+        },
+        join() {
+          return this;
+        },
+        groupBy() {
+          return this;
+        },
+        orderBy() {
+          return this;
+        },
+        limit() {
+          return this;
+        },
+        insert(payload) {
+          const rows = Array.isArray(payload) ? payload : [payload];
+          const upsert = (row, overrides = {}) => {
+            if (table === 'app_settings') {
+              const key = row.setting_key;
+              const existing = settingsStore.get(key) || {};
+              settingsStore.set(key, { ...existing, ...row, ...overrides });
+            }
+            return Promise.resolve();
+          };
+
+          return {
+            onConflict() {
+              return {
+                merge(overrides) {
+                  return Promise.all(rows.map((row) => upsert(row, overrides))).then(() => undefined);
+                }
+              };
+            }
+          };
+        }
+      };
+    };
+
+    const dbMock = jest.fn((table) => buildQuery(table));
+    dbMock.raw = jest.fn();
+    dbMock.transaction = async (handler) => handler({
+      commit: async () => {},
+      rollback: async () => {}
+    });
+
+    jest.doMock('../src/database/db', () => ({
+      db: dbMock,
+      logActivity: jest.fn()
+    }));
+
+    jest.doMock('../src/middleware/auth', () => ({
+      adminAuth: (req, res, next) => {
+        req.admin = { id: 1, username: 'tester' };
+        next();
+      }
+    }));
+
+    jest.doMock('../src/services/publicSiteService', () => ({
+      clearPublicSiteCache: jest.fn(),
+      getDefaultPublicSitePayload: jest.fn(),
+      getRawPublicSiteSettings: jest.fn().mockResolvedValue({})
+    }));
+
+    jest.doMock('../src/services/rateLimitService', () => ({
+      clearSettingsCache: jest.fn()
+    }));
+
+    jest.doMock('../src/middleware/maintenance', () => ({
+      maintenanceMiddleware: (req, res, next) => next(),
+      clearMaintenanceCache: jest.fn()
+    }));
+
+    router = require('../src/routes/adminSettings');
+
+    app = express();
+    app.use(express.json());
+    app.use('/api/admin/settings', router);
+  });
+
+  afterEach(async () => {
+    resetModules();
+    if (tmpDir) {
+      await fsPromises.rm(tmpDir, { recursive: true, force: true });
+      tmpDir = null;
+    }
+    delete process.env.STORAGE_PATH;
+  });
+
+  it('stores logo uploads under STORAGE_PATH and deletes on branding reset', async () => {
+    const fileBuffer = Buffer.from('fake image data');
+
+    const uploadResponse = await request(app)
+      .post('/api/admin/settings/logo')
+      .attach('logo', fileBuffer, 'logo.png');
+
+    expect(uploadResponse.status).toBe(200);
+    expect(uploadResponse.body).toHaveProperty('logoUrl');
+    const logoUrl = uploadResponse.body.logoUrl;
+    expect(logoUrl.startsWith('/uploads/logos/')).toBe(true);
+
+    const storedPath = path.join(tmpDir, logoUrl.replace('/uploads/', 'uploads/'));
+    await expect(fsPromises.access(storedPath)).resolves.toBeUndefined();
+
+    await request(app)
+      .put('/api/admin/settings/branding')
+      .send({
+        company_name: 'Test Co',
+        company_tagline: 'Tagline',
+        support_email: 'test@example.com',
+        footer_text: 'Footer',
+        watermark_enabled: false,
+        watermark_position: 'bottom-right',
+        watermark_opacity: 0.5,
+        watermark_size: 'medium',
+        favicon_url: null,
+        logo_url: '',
+        watermark_logo_url: null,
+        logo_size: 'medium',
+        logo_max_height: 120,
+        logo_position: 'left',
+        logo_display_header: true,
+        logo_display_hero: false,
+        logo_display_mode: 'default'
+      })
+      .expect(200);
+
+    await expect(fsPromises.access(storedPath)).rejects.toThrow();
+  });
+});

backend/__tests__/integration/adminArchives.restoreCategories.test.js

@@ -0,0 +1,498 @@
+/**
+ * Restoring an archive must put the photos back into their categories.
+ *
+ * The archive writer already persists `category_name` per photo in
+ * `photos_manifest.json` — that is why the manifest exists, and the comment
+ * above it says so: "(and category linkage) can't be derived from the
+ * extracted files alone". The restore route then read only
+ * `original_filename` from it and kept deriving the category from the ZIP's
+ * first path segment.
+ *
+ * Archives store photos exactly as they sit on disk, so an event whose photos
+ * live in the gallery root produces a FLAT zip. `path.dirname()` is '.' for
+ * every entry, no category is resolved, and every restored photo lands with
+ * `category_id = null` — silently, with a 200 response.
+ *
+ * These pin the manifest as the source of truth, with the directory as the
+ * fallback that keeps foldered and legacy archives working.
+ */
+
+const fs = require('fs');
+const path = require('path');
+const os = require('os');
+const express = require('express');
+const request = require('supertest');
+
+describe('archive restore restores categories (flat archives included)', () => {
+  let tmpDir; let db; let cleanup; let app; let storagePath;
+
+  beforeAll(async () => {
+    tmpDir = await fs.promises.mkdtemp(path.join(os.tmpdir(), 'picpeak-restore-cat-'));
+    storagePath = path.join(tmpDir, 'storage');
+    process.env.NODE_ENV = 'test';
+    process.env.TEST_DATABASE_PATH = path.join(tmpDir, 'data', 'test.db');
+    process.env.STORAGE_PATH = storagePath;
+    await fs.promises.mkdir(path.dirname(process.env.TEST_DATABASE_PATH), { recursive: true });
+    await fs.promises.mkdir(path.join(storagePath, 'archives'), { recursive: true });
+
+    jest.resetModules();
+    jest.doMock('../../src/middleware/auth', () => ({
+      adminAuth: (req, _res, next) => { req.admin = { id: 1, username: 'tester' }; next(); },
+    }));
+    jest.doMock('../../src/middleware/permissions', () => ({
+      requirePermission: () => (_req, _res, next) => next(),
+    }));
+    jest.doMock('../../src/middleware/ownership', () => ({
+      requireEventOwnership: (_req, _res, next) => next(),
+    }));
+
+    ({ db, cleanup } = await require('./helpers/crmDb').bootCrmDb());
+    // bootCrmDb points STORAGE_PATH at its own tmp dir; follow it rather than
+    // fighting it, so the archives the tests write are where the route looks.
+    storagePath = process.env.STORAGE_PATH;
+    await fs.promises.mkdir(path.join(storagePath, 'archives'), { recursive: true });
+
+    app = express();
+    app.use(express.json());
+    app.use('/admin/archives', require('../../src/routes/adminArchives'));
+  }, 180000);
+
+  afterAll(async () => {
+    if (cleanup) await cleanup();
+    await fs.promises.rm(tmpDir, { recursive: true, force: true }).catch(() => {});
+  });
+
+  beforeEach(async () => {
+    await db('photos').del();
+    await db('photo_categories').del();
+    await db('events').del();
+  });
+
+  /** A one-pixel JPEG is enough; the route only stats the extracted file. */
+  const PIXEL = Buffer.from(
+    '/9j/4AAQSkZJRgABAQEAYABgAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0a'
+    + 'HBwgJC4nICIsIxwcKDcpLDAxNDQ0Hyc5PTgyPC4zNDL/wAALCAABAAEBAREA/8QAFAABAAAAAAAA'
+    + 'AAAAAAAAAAAACf/EABQQAQAAAAAAAAAAAAAAAAAAAAD/2gAIAQEAAD8AKp//2Q==',
+    'base64',
+  );
+
+  async function writeArchive(name, entries) {
+    // Required lazily: the suite calls jest.resetModules() in beforeAll, and
+    // archiver's readable-stream copy does not survive being split across the
+    // two module registries.
+    const archiver = require('archiver');
+    const archivePath = path.join(storagePath, 'archives', name);
+    await new Promise((resolve, reject) => {
+      const output = fs.createWriteStream(archivePath);
+      const zip = archiver('zip', { zlib: { level: 0 } });
+      output.on('close', resolve);
+      zip.on('error', reject);
+      zip.pipe(output);
+      for (const [entryName, buffer] of Object.entries(entries)) {
+        zip.append(buffer, { name: entryName });
+      }
+      zip.finalize();
+    });
+    return path.join('archives', name);
+  }
+
+  async function seedArchivedEvent(archiveRelPath, slug) {
+    const [row] = await db('events').insert({
+      slug,
+      event_type: 'wedding',
+      event_name: slug,
+      event_date: '2026-06-27',
+      host_email: 'h@example.com',
+      admin_email: 'a@example.com',
+      password_hash: 'x',
+      share_link: `${slug}-share`,
+      expires_at: new Date().toISOString(),
+      is_archived: 1,  // sqlite stores booleans as 0/1, see utils/dbCompat
+      archive_path: archiveRelPath,
+    }).returning('id');
+    return typeof row === 'object' ? row.id : row;
+  }
+
+  const categoryOf = async (filename) => {
+    const photo = await db('photos').where('filename', filename).first();
+    if (!photo || !photo.category_id) return null;
+    const category = await db('photo_categories').where('id', photo.category_id).first();
+    return category ? category.name : null;
+  };
+
+  it('takes the category from the manifest when the archive is flat', async () => {
+    // Exactly the shape a gallery-root event archives to: no directories.
+    const manifest = JSON.stringify([
+      { filename: 'a.jpg', original_filename: 'DSC_0001.jpg', category_name: 'Polterabend' },
+      { filename: 'b.jpg', original_filename: 'DSC_0002.jpg', category_name: 'Ceremony' },
+    ]);
+    const archiveRelPath = await writeArchive('flat.zip', {
+      'a.jpg': PIXEL,
+      'b.jpg': PIXEL,
+      'photos_manifest.json': Buffer.from(manifest, 'utf8'),
+    });
+    const eventId = await seedArchivedEvent(archiveRelPath, 'flat-event');
+
+    const res = await request(app).post(`/admin/archives/${eventId}/restore`).send({});
+    expect(res.status).toBe(200);
+
+    // The whole bug: both of these used to be null.
+    expect(await categoryOf('a.jpg')).toBe('Polterabend');
+    expect(await categoryOf('b.jpg')).toBe('Ceremony');
+  });
+
+  it('stores a real timestamp on restored photos, not "[object Object]"', async () => {
+    // The jest+sqlite landmine: a Date handed to knex inside jest stores as
+    // the literal string "[object Object]". Production writes ms-numbers and
+    // is unaffected, so this only ever corrupts what tests read back — which
+    // is how it survives unnoticed.
+    const archiveRelPath = await writeArchive('timestamp.zip', {
+      'individual/STAMPED.jpg': PIXEL,
+      'photos_manifest.json': Buffer.from(JSON.stringify([
+        { filename: 'STAMPED.jpg', original_filename: 'STAMPED.jpg', category_name: 'Ceremony' },
+      ]), 'utf8'),
+    });
+    const eventId = await seedArchivedEvent(archiveRelPath, 'timestamp-event');
+
+    const res = await request(app).post(`/admin/archives/${eventId}/restore`).send({});
+    expect(res.status).toBe(200);
+
+    const photo = await db('photos').where({ event_id: eventId, filename: 'STAMPED.jpg' }).first();
+    expect(String(photo.uploaded_at)).not.toBe('[object Object]');
+    expect(Number.isNaN(new Date(photo.uploaded_at).getTime())).toBe(false);
+  });
+
+  it('reuses an existing category row instead of creating a duplicate', async () => {
+    const archiveRelPath = await writeArchive('reuse.zip', {
+      'c.jpg': PIXEL,
+      'photos_manifest.json': Buffer.from(JSON.stringify([
+        { filename: 'c.jpg', original_filename: 'DSC_0003.jpg', category_name: 'Party' },
+      ]), 'utf8'),
+    });
+    const eventId = await seedArchivedEvent(archiveRelPath, 'reuse-event');
+    await db('photo_categories').insert({
+      event_id: eventId, name: 'Party', slug: 'party', created_at: new Date(),
+    });
+
+    const res = await request(app).post(`/admin/archives/${eventId}/restore`).send({});
+    expect(res.status).toBe(200);
+
+    expect(await categoryOf('c.jpg')).toBe('Party');
+    const rows = await db('photo_categories').where({ event_id: eventId, name: 'Party' });
+    expect(rows).toHaveLength(1);
+  });
+
+  it('still falls back to the directory for legacy archives with no manifest', async () => {
+    // No manifest at all — the shape every archive had before the manifest
+    // landed. The directory is the only signal left, and it must keep working.
+    //
+    // `individual/` is what a REAL archive contains: entry names are the
+    // storage key minus `events/active/{slug}`, and that layout is
+    // `individual/` / `collages/`. Categories have never been directories, so
+    // the fallback invents a category with that name — not useful, but better
+    // than losing every category, and this pins what actually happens rather
+    // than a category-shaped folder no archive produces.
+    const archiveRelPath = await writeArchive('foldered.zip', {
+      'individual/d.jpg': PIXEL,
+    });
+    const eventId = await seedArchivedEvent(archiveRelPath, 'foldered-event');
+
+    const res = await request(app).post(`/admin/archives/${eventId}/restore`).send({});
+    expect(res.status).toBe(200);
+
+    expect(await categoryOf('d.jpg')).toBe('individual');
+  });
+
+  it('reuses a GLOBAL category instead of cloning it into the event', async () => {
+    // Seeded categories (Ceremony, Reception, ...) have event_id NULL. An
+    // event-only lookup misses them, so the restore used to create a second
+    // "Ceremony" — and because is_global defaults to TRUE, that duplicate then
+    // appeared in every other event's category list.
+    const [g] = await db('photo_categories').insert({
+      event_id: null, name: 'Ceremony', slug: 'ceremony', is_global: true, created_at: new Date(),
+    }).returning('id');
+    const globalId = typeof g === 'object' ? g.id : g;
+
+    const archiveRelPath = await writeArchive('global.zip', {
+      'individual/gl.jpg': PIXEL,
+      'photos_manifest.json': Buffer.from(JSON.stringify([
+        { filename: 'gl.jpg', original_filename: 'DSC_1.jpg', category_name: 'Ceremony' },
+      ]), 'utf8'),
+    });
+    const eventId = await seedArchivedEvent(archiveRelPath, 'global-event');
+
+    const res = await request(app).post(`/admin/archives/${eventId}/restore`).send({});
+    expect(res.status).toBe(200);
+
+    const photo = await db('photos').where('filename', 'gl.jpg').first();
+    expect(photo.category_id).toBe(globalId);
+    // No clone, global or otherwise.
+    const all = await db('photo_categories').where('name', 'Ceremony');
+    expect(all).toHaveLength(1);
+  });
+
+  it('does not create a GLOBAL category when it has to invent one', async () => {
+    // is_global defaults to true on this column, so an unqualified insert would
+    // leak a restore's category name into every gallery on the instance.
+    const archiveRelPath = await writeArchive('newcat.zip', {
+      'individual/nc.jpg': PIXEL,
+      'photos_manifest.json': Buffer.from(JSON.stringify([
+        { filename: 'nc.jpg', original_filename: 'DSC_2.jpg', category_name: 'Polterabend' },
+      ]), 'utf8'),
+    });
+    const eventId = await seedArchivedEvent(archiveRelPath, 'newcat-event');
+
+    const res = await request(app).post(`/admin/archives/${eventId}/restore`).send({});
+    expect(res.status).toBe(200);
+
+    const created = await db('photo_categories').where('name', 'Polterabend').first();
+    expect(created.event_id).toBe(eventId);
+    expect(created.is_global === false || created.is_global === 0).toBe(true);
+  });
+
+  it('matches the manifest when the ZIP was written with original filenames', async () => {
+    // With general_use_original_filenames_for_downloads on at archive time,
+    // archiveService names entries after the ORIGINAL filename while the
+    // manifest stays keyed by photos.filename. Looking up the extracted
+    // basename missed every entry, so categories were lost on exactly those
+    // archives.
+    const archiveRelPath = await writeArchive('original-names.zip', {
+      'individual/DSC_4242.jpg': PIXEL,
+      'photos_manifest.json': Buffer.from(JSON.stringify([
+        { filename: 'stored_9f8e7d.jpg', original_filename: 'DSC_4242.jpg', category_name: 'Drohne' },
+      ]), 'utf8'),
+    });
+    const eventId = await seedArchivedEvent(archiveRelPath, 'original-names-event');
+
+    const res = await request(app).post(`/admin/archives/${eventId}/restore`).send({});
+    expect(res.status).toBe(200);
+
+    expect(await categoryOf('DSC_4242.jpg')).toBe('Drohne');
+  });
+
+  it('prefers the event-scoped category when a global shares its name', async () => {
+    // The category API permits both. A single OR-lookup with .first() returned
+    // whichever the engine chose, so a photo could be reassigned to the global
+    // row and lose event-local settings such as allow_downloads.
+    const archiveRelPath = await writeArchive('collide.zip', {
+      'individual/co.jpg': PIXEL,
+      'photos_manifest.json': Buffer.from(JSON.stringify([
+        { filename: 'co.jpg', original_filename: 'DSC_3.jpg', category_name: 'Reception' },
+      ]), 'utf8'),
+    });
+    const eventId = await seedArchivedEvent(archiveRelPath, 'collide-event');
+
+    await db('photo_categories').insert({
+      event_id: null, name: 'Reception', slug: 'reception-global', is_global: true, created_at: new Date(),
+    });
+    const [own] = await db('photo_categories').insert({
+      event_id: eventId, name: 'Reception', slug: 'reception-own', is_global: false, created_at: new Date(),
+    }).returning('id');
+    const ownId = typeof own === 'object' ? own.id : own;
+
+    const res = await request(app).post(`/admin/archives/${eventId}/restore`).send({});
+    expect(res.status).toBe(200);
+
+    const photo = await db('photos').where('filename', 'co.jpg').first();
+    expect(photo.category_id).toBe(ownId);
+  });
+
+  it('matches a sanitized original filename, as the ZIP would have written it', async () => {
+    // archiveService runs original names through sanitizeForZipEntry() before
+    // writing the entry, so the emitted name differs from the manifest column.
+    const archiveRelPath = await writeArchive('sanitized.zip', {
+      'individual/od_dr_DSC_5.jpg': PIXEL,
+      'photos_manifest.json': Buffer.from(JSON.stringify([
+        { filename: 'stored_abc.jpg', original_filename: 'od/dr/DSC_5.jpg', category_name: 'Strand' },
+      ]), 'utf8'),
+    });
+    const eventId = await seedArchivedEvent(archiveRelPath, 'sanitized-event');
+
+    const res = await request(app).post(`/admin/archives/${eventId}/restore`).send({});
+    expect(res.status).toBe(200);
+
+    expect(await categoryOf('od_dr_DSC_5.jpg')).toBe('Strand');
+  });
+
+  it('ignores a legacy event-owned row when falling back to globals', async () => {
+    // The bug fixed here left rows behind on upgraded instances: event-owned
+    // AND is_global true, because the column defaults true. Matching on the
+    // flag alone would let one event's leftover be adopted by another event's
+    // restore, tying photos to a category that vanishes with someone else's
+    // gallery.
+    const otherEventId = await seedArchivedEvent('archives/none.zip', 'legacy-owner-event');
+    await db('photo_categories').insert({
+      event_id: otherEventId, name: 'Sunset', slug: 'sunset-legacy',
+      is_global: true, created_at: new Date(),
+    });
+
+    const archiveRelPath = await writeArchive('legacy-global.zip', {
+      'individual/lg.jpg': PIXEL,
+      'photos_manifest.json': Buffer.from(JSON.stringify([
+        { filename: 'lg.jpg', original_filename: 'DSC_6.jpg', category_name: 'Sunset' },
+      ]), 'utf8'),
+    });
+    const eventId = await seedArchivedEvent(archiveRelPath, 'legacy-global-event');
+
+    const res = await request(app).post(`/admin/archives/${eventId}/restore`).send({});
+    expect(res.status).toBe(200);
+
+    const photo = await db('photos').where('filename', 'lg.jpg').first();
+    const cat = await db('photo_categories').where('id', photo.category_id).first();
+    // Its own row, not the other event's leftover.
+    expect(cat.event_id).toBe(eventId);
+  });
+
+  it('drops an ambiguous original-name alias rather than guessing', async () => {
+    // Two photos in different ZIP folders can share an original basename;
+    // archiveService treats the paths as distinct and suffixes neither. Both
+    // would collapse onto one alias, and whichever won would hand the other
+    // photo someone else's category.
+    const archiveRelPath = await writeArchive('ambiguous.zip', {
+      'individual/SHARED.jpg': PIXEL,
+      'photos_manifest.json': Buffer.from(JSON.stringify([
+        { filename: 'a_stored.jpg', original_filename: 'SHARED.jpg', category_name: 'Alpha' },
+        { filename: 'b_stored.jpg', original_filename: 'SHARED.jpg', category_name: 'Beta' },
+      ]), 'utf8'),
+    });
+    const eventId = await seedArchivedEvent(archiveRelPath, 'ambiguous-event');
+
+    const res = await request(app).post(`/admin/archives/${eventId}/restore`).send({});
+    expect(res.status).toBe(200);
+
+    // Falls back to the directory rather than picking Alpha or Beta at random.
+    expect(await categoryOf('SHARED.jpg')).toBe('individual');
+    for (const name of ['Alpha', 'Beta']) {
+      expect(await db('photo_categories').where({ event_id: eventId, name }).first()).toBeFalsy();
+    }
+  });
+
+  it('honours a manifest that says UNCATEGORIZED, instead of inventing one from the directory', async () => {
+    // The case the manifest-first change was for. A real archive puts every
+    // photo under `individual/`, so a photo the manifest records as having no
+    // category used to come back filed under a category called "individual" —
+    // the manifest being authoritative for "category X" but not for "none".
+    const manifest = JSON.stringify([
+      { filename: 'u.jpg', original_filename: 'DSC_7000.jpg', category_name: null },
+    ]);
+    const archiveRelPath = await writeArchive('uncategorized.zip', {
+      'individual/u.jpg': PIXEL,
+      'photos_manifest.json': Buffer.from(manifest, 'utf8'),
+    });
+    const eventId = await seedArchivedEvent(archiveRelPath, 'uncategorized-event');
+
+    const res = await request(app).post(`/admin/archives/${eventId}/restore`).send({});
+    expect(res.status).toBe(200);
+
+    expect(await categoryOf('u.jpg')).toBeNull();
+    // And no junk category row was created as a side effect.
+    const rows = await db('photo_categories').where({ event_id: eventId });
+    expect(rows).toHaveLength(0);
+  });
+
+  it('drops a canonical filename that two photos claim, rather than guessing', async () => {
+    // photos.filename is not unique within an event: s3AutoImporter takes
+    // path.basename(entry.key) and dedupes by path, so two imported files in
+    // different subfolders both land as IMG_1234.jpg. Both ZIP entries reduce
+    // to the same basename at restore, so keeping the last row seen would give
+    // one photo the other's category.
+    const archiveRelPath = await writeArchive('dup-canonical.zip', {
+      'individual/IMG_1234.jpg': PIXEL,
+      'photos_manifest.json': Buffer.from(JSON.stringify([
+        { filename: 'IMG_1234.jpg', original_filename: 'a.jpg', category_name: 'Alpha' },
+        { filename: 'IMG_1234.jpg', original_filename: 'b.jpg', category_name: 'Beta' },
+      ]), 'utf8'),
+    });
+    const eventId = await seedArchivedEvent(archiveRelPath, 'dup-canonical-event');
+
+    const res = await request(app).post(`/admin/archives/${eventId}/restore`).send({});
+    expect(res.status).toBe(200);
+
+    expect(await categoryOf('IMG_1234.jpg')).toBe('individual');
+    for (const name of ['Alpha', 'Beta']) {
+      expect(await db('photo_categories').where({ event_id: eventId, name }).first()).toBeFalsy();
+    }
+  });
+
+  it("drops a name that one row owns canonically and another claims as an alias", async () => {
+    // Undecidable: with original-filename archiving ON the ZIP entry under
+    // this name is the ALIAS owner's file, with it OFF it is the canonical
+    // owner's, and the manifest does not record which mode was used. The
+    // point of the two-pass split is that this now resolves the same way
+    // every run — the archive query has no ORDER BY, so it used to be a coin
+    // flip between dropping the name and overwriting it.
+    const archiveRelPath = await writeArchive('alias-vs-canonical.zip', {
+      'individual/CANON.jpg': PIXEL,
+      'photos_manifest.json': Buffer.from(JSON.stringify([
+        { filename: 'CANON.jpg', original_filename: 'unrelated.jpg', category_name: 'Canonical' },
+        { filename: 'other_stored.jpg', original_filename: 'CANON.jpg', category_name: 'Aliased' },
+      ]), 'utf8'),
+    });
+    const eventId = await seedArchivedEvent(archiveRelPath, 'alias-vs-canonical-event');
+
+    const res = await request(app).post(`/admin/archives/${eventId}/restore`).send({});
+    expect(res.status).toBe(200);
+
+    // Falls back to the directory rather than guessing either row.
+    expect(await categoryOf('CANON.jpg')).toBe('individual');
+    for (const name of ['Canonical', 'Aliased']) {
+      expect(await db('photo_categories').where({ event_id: eventId, name }).first()).toBeFalsy();
+    }
+  });
+
+  it('picks the lowest id and warns when two categories share a name', async () => {
+    // Allowed: two event-scoped categories with the same display name and
+    // different slugs. .first() used to pick either, so a re-run could move
+    // photos between them and inherit the wrong allow_downloads.
+    const archiveRelPath = await writeArchive('dupe-category.zip', {
+      'individual/DUPE.jpg': PIXEL,
+      'photos_manifest.json': Buffer.from(JSON.stringify([
+        { filename: 'DUPE.jpg', original_filename: 'DUPE.jpg', category_name: 'Ceremony' },
+      ]), 'utf8'),
+    });
+    const eventId = await seedArchivedEvent(archiveRelPath, 'dupe-category-event');
+
+    const [first] = await db('photo_categories').insert({
+      name: 'Ceremony', slug: 'ceremony-a', is_global: 0, event_id: eventId,
+    }).returning('id');
+    await db('photo_categories').insert({
+      name: 'Ceremony', slug: 'ceremony-b', is_global: 0, event_id: eventId,
+    });
+    const firstId = typeof first === 'object' ? first.id : first;
+
+    const res = await request(app).post(`/admin/archives/${eventId}/restore`).send({});
+    expect(res.status).toBe(200);
+
+    // Stable, not arbitrary: the same run twice lands on the same row.
+    const photo = await db('photos').where({ event_id: eventId, filename: 'DUPE.jpg' }).first();
+    expect(photo.category_id).toBe(firstId);
+    // And no third "Ceremony" was invented.
+    expect((await db('photo_categories').where({ event_id: eventId, name: 'Ceremony' })).length)
+      .toBe(2);
+  });
+
+  it('does not invent a category for a photo row that already exists', async () => {
+    // archiveEvent retains photo rows, so a restore can skip every insert.
+    // Resolving categories before that check created one from the stale
+    // manifest name that nothing then used — renaming a category while its
+    // event was archived left the old name behind as an empty duplicate.
+    const archiveRelPath = await writeArchive('existing-rows.zip', {
+      'individual/KEPT.jpg': PIXEL,
+      'photos_manifest.json': Buffer.from(JSON.stringify([
+        { filename: 'KEPT.jpg', original_filename: 'KEPT.jpg', category_name: 'OldName' },
+      ]), 'utf8'),
+    });
+    const eventId = await seedArchivedEvent(archiveRelPath, 'existing-rows-event');
+    await db('photos').insert({
+      event_id: eventId, filename: 'KEPT.jpg', path: 'whatever/KEPT.jpg', type: 'jpg',
+      uploaded_at: new Date().toISOString(),
+    });
+
+    const res = await request(app).post(`/admin/archives/${eventId}/restore`).send({});
+    expect(res.status).toBe(200);
+
+    expect(await db('photo_categories').where({ event_id: eventId, name: 'OldName' }).first())
+      .toBeFalsy();
+  });
+
+});

backend/__tests__/integration/adminBackupCoverage.test.js

@@ -0,0 +1,242 @@
+/**
+ * Integration test for GET /api/admin/system-health/backup-coverage.
+ *
+ * Pins the Stage C diagnostic that tells admins what the next
+ * "Run Backup Now" will include, skip, or silently miss.
+ *
+ * Test surface:
+ *   1. Empty / fresh install → default seed (7 paths), inline mode,
+ *      no DB dump on file yet, no drift
+ *   2. Toggle `include_in_default=false` → coverage flips to
+ *      'skipped-by-toggle'
+ *   3. Feature_flag gating reflects the actual app_settings value
+ *      (events/archived ⇄ backup_include_archived)
+ *   4. Drift detection: a top-level subdir on disk with no
+ *      `backup_paths` row is flagged in `unconfiguredOnDisk`
+ *   5. Allow-list: `backups/` and `tmp/` are never flagged as drift
+ *   6. Scheduled-only mode + recent dump → `database.ok = true`
+ *   7. Scheduled-only mode + stale (>26h) dump → `database.ok = false`
+ *      and `lastDumpStale = true`
+ *
+ * Same auth/permission pass-through strategy as
+ * adminBackupIntegrity.test.js — we exercise the route's logic,
+ * not the auth middleware.
+ */
+
+const fs = require('fs');
+const path = require('path');
+const express = require('express');
+const request = require('supertest');
+
+const { bootCrmDb } = require('./helpers/crmDb');
+
+jest.mock('../../src/middleware/auth', () => ({
+  adminAuth: (req, _res, next) => { req.admin = { id: 1 }; next(); },
+  customerAuth: (_req, _res, next) => next(),
+  galleryAuth: (_req, _res, next) => next(),
+}));
+
+jest.mock('../../src/middleware/permissions', () => ({
+  requirePermission: () => (_req, _res, next) => next(),
+}));
+
+jest.setTimeout(120000);
+
+describe('GET /api/admin/system-health/backup-coverage', () => {
+  let db;
+  let cleanup;
+  let storagePath;
+  let app;
+
+  beforeAll(async () => {
+    ({ db, cleanup } = await bootCrmDb());
+    storagePath = process.env.STORAGE_PATH;
+
+    const route = require('../../src/routes/adminSystemHealth');
+    app = express();
+    app.use(express.json());
+    app.use('/api/admin/system-health', route);
+  }, 120000);
+
+  afterAll(async () => {
+    if (cleanup) await cleanup();
+  });
+
+  function mkdir(rel) {
+    fs.mkdirSync(path.join(storagePath, rel), { recursive: true });
+  }
+
+  function rmdir(rel) {
+    fs.rmSync(path.join(storagePath, rel), { recursive: true, force: true });
+  }
+
+  async function restoreDefaultPaths() {
+    await db('backup_paths').del();
+    const { DEFAULT_PATHS } = require('../../migrations/core/109_add_backup_paths');
+    await db('backup_paths').insert(DEFAULT_PATHS.map((row) => ({
+      ...row,
+      created_at: new Date(),
+      updated_at: new Date(),
+    })));
+  }
+
+  beforeEach(async () => {
+    await restoreDefaultPaths();
+    await db('database_backup_runs').del().catch(() => {});
+    await db('app_settings').where('setting_type', 'backup').del().catch(() => {});
+  });
+
+  it('returns the canonical 7 paths + database block on a fresh install', async () => {
+    const res = await request(app).get('/api/admin/system-health/backup-coverage');
+    expect(res.status).toBe(200);
+    expect(res.body).toHaveProperty('report');
+
+    const { report } = res.body;
+    expect(report.paths.map((p) => p.path)).toEqual([
+      'events/active',
+      'events/archived',
+      'thumbnails',
+      'previews',
+      'heroes',
+      'uploads',
+      'business-docs',
+    ]);
+
+    // Default mode is inline — no inline_dump setting present means
+    // "inline is ON" (matches ensureDatabaseDumpForBackup semantics).
+    expect(report.database.mode).toBe('inline');
+    expect(report.database.ok).toBe(true);
+
+    expect(report.summary).toMatchObject({
+      configuredCount: 7,
+      tableMissingFallbackInUse: false,
+    });
+  });
+
+  it('flips a path to skipped-by-toggle when include_in_default=false', async () => {
+    await db('backup_paths').where('path', 'thumbnails').update({
+      include_in_default: false,
+    });
+
+    const res = await request(app).get('/api/admin/system-health/backup-coverage');
+    const thumbnails = res.body.report.paths.find((p) => p.path === 'thumbnails');
+    expect(thumbnails.coverage).toBe('skipped-by-toggle');
+    expect(thumbnails.includeInDefault).toBe(false);
+  });
+
+  it('feature_flag gating reflects app_settings (archived path off vs on)', async () => {
+    // backup_include_archived not set → archived skipped via flag
+    const off = await request(app).get('/api/admin/system-health/backup-coverage');
+    const archivedOff = off.body.report.paths.find((p) => p.path === 'events/archived');
+    expect(archivedOff.coverage).toBe('skipped-by-feature-flag');
+    expect(archivedOff.featureFlag).toBe('backup_include_archived');
+    expect(archivedOff.featureFlagValue).toBe(null); // unset
+
+    // Now set the flag — but path is missing on disk, so coverage
+    // resolves to 'missing-on-disk', proving the flag was honoured.
+    await db('app_settings').insert({
+      setting_key: 'backup_include_archived',
+      setting_value: JSON.stringify(true),
+      setting_type: 'backup',
+    }).onConflict('setting_key').merge();
+
+    const on = await request(app).get('/api/admin/system-health/backup-coverage');
+    const archivedOn = on.body.report.paths.find((p) => p.path === 'events/archived');
+    expect(archivedOn.featureFlagValue).toBe(true);
+    // No on-disk dir → 'missing-on-disk' (not 'skipped-by-feature-flag')
+    expect(['missing-on-disk', 'will-scan']).toContain(archivedOn.coverage);
+  });
+
+  it('detects unconfigured top-level subdirs as drift', async () => {
+    mkdir('events/active');           // configured
+    mkdir('plugin-store/cache');      // DRIFT
+    mkdir('shiny-new-feature/data');  // DRIFT
+
+    const res = await request(app).get('/api/admin/system-health/backup-coverage');
+    expect(res.body.report.drift.unconfiguredOnDisk).toEqual(expect.arrayContaining([
+      'plugin-store',
+      'shiny-new-feature',
+    ]));
+    expect(res.body.report.drift.unconfiguredOnDisk).not.toContain('events');
+
+    rmdir('plugin-store');
+    rmdir('shiny-new-feature');
+  });
+
+  it('never flags backups/ or tmp/ as drift (allow-list)', async () => {
+    mkdir('backups');
+    mkdir('tmp');
+
+    const res = await request(app).get('/api/admin/system-health/backup-coverage');
+    expect(res.body.report.drift.unconfiguredOnDisk).not.toContain('backups');
+    expect(res.body.report.drift.unconfiguredOnDisk).not.toContain('tmp');
+    expect(res.body.report.drift.expectedNonBackupDirs).toEqual(
+      expect.arrayContaining(['backups', 'tmp']),
+    );
+
+    rmdir('backups');
+    rmdir('tmp');
+  });
+
+  it('scheduled-only mode + recent dump → database.ok=true, not stale', async () => {
+    await db('app_settings').insert({
+      setting_key: 'backup_database_inline_dump',
+      setting_value: JSON.stringify(false),
+      setting_type: 'backup',
+    }).onConflict('setting_key').merge();
+
+    const recentDump = path.join(storagePath, 'backups', 'recent.sql.gz');
+    fs.mkdirSync(path.dirname(recentDump), { recursive: true });
+    fs.writeFileSync(recentDump, 'pretend dump');
+    await db('database_backup_runs').insert({
+      started_at: new Date(),
+      completed_at: new Date(),  // just now
+      status: 'completed',
+      backup_type: 'pg',
+      file_path: recentDump,
+      file_size_bytes: fs.statSync(recentDump).size,
+      destination_path: recentDump,
+    });
+
+    const res = await request(app).get('/api/admin/system-health/backup-coverage');
+    expect(res.body.report.database.mode).toBe('scheduled-only');
+    expect(res.body.report.database.inlineDumpExplicitlyDisabled).toBe(true);
+    expect(res.body.report.database.lastDumpStale).toBe(false);
+    expect(res.body.report.database.ok).toBe(true);
+  });
+
+  it('scheduled-only mode + stale dump → database.ok=false, lastDumpStale=true', async () => {
+    await db('app_settings').insert({
+      setting_key: 'backup_database_inline_dump',
+      setting_value: JSON.stringify(false),
+      setting_type: 'backup',
+    }).onConflict('setting_key').merge();
+
+    const oldDump = path.join(storagePath, 'backups', 'old.sql.gz');
+    fs.mkdirSync(path.dirname(oldDump), { recursive: true });
+    fs.writeFileSync(oldDump, 'pretend old dump');
+    // 48 hours ago — well past the 26h staleness threshold. ISO
+    // string instead of a Date object because knex-sqlite's datetime
+    // serialisation has a quirk where some Date instances coerce to
+    // '[object Object]' on insert (the test 6 "recent dump" case
+    // passes only because `new Date()` happens to round-trip safely;
+    // arithmetic Dates don't).
+    const stale = new Date(Date.now() - 48 * 60 * 60 * 1000).toISOString();
+    await db('database_backup_runs').insert({
+      started_at: stale,
+      completed_at: stale,
+      status: 'completed',
+      backup_type: 'pg',
+      file_path: oldDump,
+      file_size_bytes: fs.statSync(oldDump).size,
+      destination_path: oldDump,
+    });
+
+    const res = await request(app).get('/api/admin/system-health/backup-coverage');
+    expect(res.body.report.database.lastDumpStale).toBe(true);
+    expect(res.body.report.database.ok).toBe(false);
+    // Top-level summary reflects the failed DB check.
+    expect(res.body.report.summary.databaseOk).toBe(false);
+    expect(res.body.report.summary.overallOk).toBe(false);
+  });
+});

backend/__tests__/integration/adminBackupIntegrity.test.js

@@ -0,0 +1,140 @@
+/**
+ * Integration test for GET /api/admin/system-health/backup-integrity.
+ *
+ * Auth + permission middleware are mocked to pass-through so the test
+ * focuses on the route's own behaviour: scope-param validation, the
+ * successResponse envelope, and that the underlying service report
+ * surfaces correctly in the JSON body.
+ *
+ * The verifier service itself is exercised against the real schema
+ * (bootCrmDb) and real filesystem — only the auth gate is stubbed.
+ */
+
+const fs = require('fs');
+const path = require('path');
+const express = require('express');
+const request = require('supertest');
+
+const { bootCrmDb, seedMinimal } = require('./helpers/crmDb');
+
+// Pass-through auth so we don't need to mint JWTs.
+jest.mock('../../src/middleware/auth', () => ({
+  adminAuth: (req, _res, next) => { req.admin = { id: 1 }; next(); },
+  customerAuth: (_req, _res, next) => next(),
+  galleryAuth: (_req, _res, next) => next(),
+}));
+
+// Pass-through permissions so settings.view always allows.
+jest.mock('../../src/middleware/permissions', () => ({
+  requirePermission: () => (_req, _res, next) => next(),
+}));
+
+jest.setTimeout(120000);
+
+describe('GET /api/admin/system-health/backup-integrity', () => {
+  let cleanup;
+  let db;
+  let customerId;
+  let app;
+  let storagePath;
+
+  beforeAll(async () => {
+    ({ db, cleanup } = await bootCrmDb());
+    ({ customerId } = await seedMinimal(db));
+    storagePath = process.env.STORAGE_PATH;
+
+    // Mount the route on a minimal Express app. Cold-require after
+    // bootCrmDb so the route's downstream `require('../database/db')`
+    // sees the same db instance.
+    const route = require('../../src/routes/adminSystemHealth');
+    app = express();
+    app.use(express.json());
+    app.use('/api/admin/system-health', route);
+  }, 120000);
+
+  afterAll(async () => {
+    if (cleanup) await cleanup();
+  });
+
+  beforeEach(async () => {
+    await db('contracts').del().catch(() => {});
+    await db('invoices').del().catch(() => {});
+    await db('quotes').del().catch(() => {});
+  });
+
+  it('returns a report envelope when nothing references any path', async () => {
+    const res = await request(app).get('/api/admin/system-health/backup-integrity');
+    expect(res.status).toBe(200);
+    expect(res.body).toHaveProperty('report');
+    expect(res.body.report.summary).toMatchObject({
+      totalRows: 0,
+      missingFiles: 0,
+      hashMismatches: 0,
+      verifiedOk: 0,
+      existsButNoHash: 0,
+    });
+    expect(res.body.report.scopes).toEqual(expect.arrayContaining([
+      'quote', 'contract', 'contract-signature', 'invoice',
+    ]));
+  });
+
+  it('surfaces a missing file in the response payload', async () => {
+    await db('contracts').insert({
+      customer_account_id: customerId,
+      contract_number: 'C-B7-MISSING',
+      status: 'sent',
+      issue_date: '2026-01-01',
+      signed_pdf_path: 'business-docs/contract/2026/C-B7-MISSING.pdf',
+      created_at: new Date(),
+    });
+
+    const res = await request(app).get('/api/admin/system-health/backup-integrity');
+    expect(res.status).toBe(200);
+    expect(res.body.report.summary.missingFiles).toBe(1);
+    expect(res.body.report.missing[0]).toMatchObject({
+      table: 'contracts',
+      column: 'signed_pdf_path',
+      expectedPath: 'business-docs/contract/2026/C-B7-MISSING.pdf',
+    });
+  });
+
+  it('honours the ?scope=invoice filter', async () => {
+    // Seed both an invoice and a contract with missing files. With
+    // scope=invoice the contract row must not appear.
+    await db('invoices').insert({
+      customer_account_id: customerId,
+      invoice_number: 'INV-B7-SCOPE',
+      status: 'sent',
+      issue_date: '2026-01-01',
+      due_date: '2026-01-31',
+      pdf_path: 'business-docs/invoice/2026/INV-B7-SCOPE.pdf',
+      created_at: new Date(),
+    });
+    await db('contracts').insert({
+      customer_account_id: customerId,
+      contract_number: 'C-B7-SCOPE',
+      status: 'sent',
+      issue_date: '2026-01-01',
+      signed_pdf_path: 'business-docs/contract/2026/C-B7-SCOPE.pdf',
+      created_at: new Date(),
+    });
+
+    const res = await request(app)
+      .get('/api/admin/system-health/backup-integrity')
+      .query({ scope: 'invoice' });
+    expect(res.status).toBe(200);
+    expect(res.body.report.scopes).toEqual(['invoice']);
+    expect(res.body.report.missing.every((m) => m.table === 'invoices')).toBe(true);
+  });
+
+  it('rejects an unknown scope with 400 + a code', async () => {
+    const res = await request(app)
+      .get('/api/admin/system-health/backup-integrity')
+      .query({ scope: 'gallery' });
+    expect(res.status).toBe(400);
+    expect(res.body.code).toBe('BACKUP_INTEGRITY_UNKNOWN_SCOPE');
+    expect(res.body.validScopes).toEqual(expect.arrayContaining([
+      'quote', 'contract', 'contract-signature', 'invoice',
+    ]));
+  });
+});

backend/__tests__/integration/adminPhotos.reference.test.js

@@ -0,0 +1,207 @@
+const fs = require('fs');
+const path = require('path');
+const os = require('os');
+const express = require('express');
+const request = require('supertest');
+
+describe('Admin photos in reference mode', () => {
+  let tmpDir;
+  let storagePath;
+  let db;
+  let app;
+  let categoryId;
+
+  const resetModules = () => {
+    jest.resetModules();
+    jest.clearAllMocks();
+  };
+
+  beforeAll(async () => {
+    tmpDir = await fs.promises.mkdtemp(path.join(os.tmpdir(), 'picpeak-admin-photos-'));
+    storagePath = path.join(tmpDir, 'storage');
+    await fs.promises.mkdir(storagePath, { recursive: true });
+
+    process.env.NODE_ENV = 'test';
+    process.env.TEST_DATABASE_PATH = path.join(tmpDir, 'data', 'photo_sharing_test.db');
+    await fs.promises.mkdir(path.dirname(process.env.TEST_DATABASE_PATH), { recursive: true });
+    try {
+      await fs.promises.unlink(process.env.TEST_DATABASE_PATH);
+    } catch (_) {
+      /* ignore */
+    }
+    process.env.STORAGE_PATH = storagePath;
+
+    resetModules();
+
+    jest.doMock('../../src/middleware/auth', () => ({
+      adminAuth: (req, _res, next) => {
+        req.admin = { id: 1, username: 'tester' };
+        next();
+      }
+    }));
+
+    jest.doMock('../../src/services/imageProcessor', () => ({
+      generateThumbnail: jest.fn().mockResolvedValue('thumbnails/mock-thumb.jpg'),
+      ensureThumbnail: jest.fn()
+    }));
+
+    jest.doMock('../../src/middleware/uploadValidation', () => ({
+      validateUploadedFiles: (_req, _res, next) => next()
+    }));
+
+    jest.doMock('../../src/utils/fileSecurityUtils', () => {
+      const actual = jest.requireActual('../../src/utils/fileSecurityUtils');
+      return {
+        ...actual,
+        validateFileType: () => true,
+        createFileUploadValidator: () => (_req, _res, next) => next()
+      };
+    });
+
+    jest.doMock('../../src/utils/logger', () => ({
+      debug: jest.fn(),
+      info: jest.fn(),
+      warn: jest.fn(),
+      error: jest.fn()
+    }));
+
+    const dbModule = require('../../src/database/db');
+    db = dbModule.db;
+
+    await db.schema.dropTableIfExists('photo_feedback');
+    await db.schema.dropTableIfExists('photos');
+    await db.schema.dropTableIfExists('photo_categories');
+    await db.schema.dropTableIfExists('events');
+
+    await db.schema.createTable('events', (table) => {
+      table.increments('id').primary();
+      table.string('slug').notNullable();
+      table.string('event_name').notNullable();
+      table.string('source_mode').notNullable();
+      table.string('external_path');
+    });
+
+    await db.schema.createTable('photo_categories', (table) => {
+      table.increments('id').primary();
+      table.string('name').notNullable();
+      table.string('slug').notNullable();
+      table.boolean('is_global').defaultTo(true);
+      table.integer('event_id');
+    });
+
+    await db.schema.createTable('photos', (table) => {
+      table.increments('id').primary();
+      table.integer('event_id').notNullable();
+      table.string('filename').notNullable();
+      table.string('path').notNullable();
+      table.string('thumbnail_path');
+      table.string('type').notNullable();
+      table.integer('size_bytes');
+      table.integer('category_id');
+      table.string('source_origin');
+      table.string('external_relpath');
+      table.datetime('uploaded_at').defaultTo(db.fn.now());
+      table.float('average_rating').defaultTo(0);
+      table.integer('like_count').defaultTo(0);
+      table.integer('favorite_count').defaultTo(0);
+    });
+
+    await db.schema.createTable('photo_feedback', (table) => {
+      table.increments('id');
+      table.integer('photo_id');
+      table.string('feedback_type');
+      table.boolean('is_approved');
+      table.boolean('is_hidden');
+    });
+
+    await db('events').insert({
+      id: 1,
+      slug: 'test-event',
+      event_name: 'Test Event',
+      source_mode: 'reference',
+      external_path: 'external/library'
+    });
+
+    const insertedCategory = await db('photo_categories').insert({
+      name: 'Highlights',
+      slug: 'highlights',
+      is_global: true
+    });
+    categoryId = Array.isArray(insertedCategory) ? insertedCategory[0] : insertedCategory;
+
+    const router = require('../../src/routes/adminPhotos');
+    app = express();
+    app.use(express.json());
+    app.use('/api/admin/events', router);
+  });
+
+  afterAll(async () => {
+    if (db) {
+      await db.destroy();
+    }
+    resetModules();
+    delete process.env.TEST_DATABASE_PATH;
+    delete process.env.STORAGE_PATH;
+    if (tmpDir) {
+      await fs.promises.rm(tmpDir, { recursive: true, force: true });
+    }
+  });
+
+  it('stores managed uploads with category information and managed origin', async () => {
+    const uploadResponse = await request(app)
+      .post(`/api/admin/events/1/upload`)
+      .field('category_id', String(categoryId))
+      .attach('photos', Buffer.from('fake image data'), 'photo.jpg');
+
+    expect(uploadResponse.status).toBe(200);
+    expect(uploadResponse.body).toHaveProperty('photos');
+    expect(Array.isArray(uploadResponse.body.photos)).toBe(true);
+
+    const photo = await db('photos').first();
+    expect(photo).toBeTruthy();
+    expect(photo.category_id).toBe(categoryId);
+    expect(photo.source_origin).toBe('managed');
+    expect(photo.external_relpath).toBeNull();
+  });
+
+  it('returns numeric category metadata when listing photos', async () => {
+    await db('photos').insert({
+      event_id: 1,
+      filename: 'external.jpg',
+      path: 'test-event/external.jpg',
+      thumbnail_path: null,
+      type: 'individual',
+      size_bytes: 123,
+      source_origin: 'external',
+      external_relpath: 'individual/external.jpg'
+    });
+
+    const response = await request(app)
+      .get(`/api/admin/events/1/photos`)
+      .expect(200);
+
+    expect(Array.isArray(response.body.photos)).toBe(true);
+    const managedPhoto = response.body.photos.find((p) => p.category_id === categoryId);
+    expect(managedPhoto).toBeTruthy();
+    expect(managedPhoto.category_name).toBe('Highlights');
+
+    const filtered = await request(app)
+      .get(`/api/admin/events/1/photos`)
+      .query({ category_id: String(categoryId) })
+      .expect(200);
+
+    expect(filtered.body.photos.every((p) => p.category_id === categoryId)).toBe(true);
+  });
+
+  it('normalizes category updates', async () => {
+    const photo = await db('photos').first();
+
+    await request(app)
+      .patch(`/api/admin/events/1/photos/${photo.id}`)
+      .send({ category_id: '0' })
+      .expect(200);
+
+    const updated = await db('photos').where({ id: photo.id }).first();
+    expect(updated.category_id).toBeNull();
+  });
+});

backend/__tests__/integration/adminThumbnails.regenerate.test.js

@@ -0,0 +1,246 @@
+/**
+ * POST /admin/thumbnails/regenerate for external/reference photos (#1129).
+ *
+ * STABLE TWIN. Diverges from the main version in one place: stable has no
+ * responsive ?w= tiers (#1095/#1109), so there is no deleteThumbnailTiers call
+ * to assert and the "drops the tiers first" test is absent here. Everything
+ * else — the external rebuild, the thumbnail_path:null contract, video
+ * skipping, per-event scoping and the superseded-key deletion — is identical.
+ *
+ * The route used to resolve every source as `storage/events/active/<path>` and
+ * `fs.access` it. External and reference rows do not live there — their
+ * originals sit under `events.external_path` — so every one of them failed the
+ * check and was counted as an error.
+ *
+ * That alone would be inert. What made it destructive is that the tier
+ * deletion runs FIRST (deliberately, so S3 and external rows are not skipped):
+ * on a reference install the button dropped every ?w= tier and rebuilt
+ * nothing, while the UI reported success — the response is sent before the
+ * background loop starts.
+ *
+ * The background work is fired with setImmediate, so every assertion here has
+ * to wait for it to drain rather than trusting the response.
+ */
+
+const fs = require('fs');
+const path = require('path');
+const os = require('os');
+const express = require('express');
+const request = require('supertest');
+
+describe('admin thumbnail regeneration (#1129)', () => {
+  let tmpDir; let db; let cleanup; let app; let imageProcessor; let storage;
+
+  beforeAll(async () => {
+    tmpDir = await fs.promises.mkdtemp(path.join(os.tmpdir(), 'picpeak-regen-'));
+    process.env.NODE_ENV = 'test';
+    process.env.TEST_DATABASE_PATH = path.join(tmpDir, 'data', 'test.db');
+    process.env.STORAGE_PATH = path.join(tmpDir, 'storage');
+    await fs.promises.mkdir(path.dirname(process.env.TEST_DATABASE_PATH), { recursive: true });
+    await fs.promises.mkdir(process.env.STORAGE_PATH, { recursive: true });
+
+    jest.resetModules();
+
+    jest.doMock('../../src/middleware/auth', () => ({
+      adminAuth: (req, _res, next) => { req.admin = { id: 1, username: 'tester' }; next(); },
+    }));
+    jest.doMock('../../src/middleware/permissions', () => ({
+      requirePermission: () => (_req, _res, next) => next(),
+    }));
+    // One instance, not a fresh object per call — the route and the
+    // assertions have to be looking at the same mock.
+    jest.doMock('../../src/services/storage', () => {
+      const instance = { delete: jest.fn().mockResolvedValue(undefined) };
+      return { getStorage: () => instance };
+    });
+    jest.doMock('../../src/services/imageProcessor', () => ({
+      ensureThumbnail: jest.fn().mockResolvedValue('thumbnails/thumb_ext1_shot.jpg'),
+      ensurePreviewImage: jest.fn().mockResolvedValue('previews/p.jpg'),
+      deletePreviewTiers: jest.fn().mockResolvedValue(undefined),
+    }));
+
+    // bootCrmDb, not run-migrations: the latter calls process.exit(0) on
+    // success, which ends the jest worker mid-suite.
+    ({ db, cleanup } = await require('./helpers/crmDb').bootCrmDb());
+
+    imageProcessor = require('../../src/services/imageProcessor');
+    storage = require('../../src/services/storage').getStorage();
+    app = express();
+    app.use(express.json());
+    app.use('/admin/thumbnails', require('../../src/routes/adminThumbnails'));
+  }, 180000);
+
+  afterAll(async () => {
+    if (cleanup) await cleanup();
+    await fs.promises.rm(tmpDir, { recursive: true, force: true }).catch(() => {});
+  });
+
+  beforeEach(async () => {
+    jest.clearAllMocks();
+    await db('photos').del();
+    await db('events').del();
+  });
+
+  async function seedEvent() {
+    const [row] = await db('events').insert({
+      slug: 'nas-wedding', event_type: 'wedding', event_name: 'nas',
+      event_date: '2026-01-01', host_email: 'h@example.com', admin_email: 'a@example.com',
+      password_hash: 'x', share_link: 'nas-share', expires_at: new Date().toISOString(),
+      source_mode: 'reference', external_path: 'weddings/2026-08',
+    }).returning('id');
+    return typeof row === 'object' ? row.id : row;
+  }
+
+  async function seedPhoto(eventId, overrides = {}) {
+    const [row] = await db('photos').insert({
+      event_id: eventId, filename: 'shot.jpg', path: 'nas-wedding/shot.jpg',
+      type: 'individual', ...overrides,
+    }).returning('id');
+    return typeof row === 'object' ? row.id : row;
+  }
+
+  /** The work runs in setImmediate; give it room to finish. */
+  const drain = () => new Promise((resolve) => setTimeout(resolve, 150));
+
+  it('rebuilds the canonical thumbnail for an external photo instead of erroring', async () => {
+    const eventId = await seedEvent();
+    await seedPhoto(eventId, {
+      source_origin: 'external',
+      external_relpath: 'shot.jpg',
+      thumbnail_path: 'thumbnails/stale.jpg',
+    });
+
+    const res = await request(app).post('/admin/thumbnails/regenerate').send({});
+    expect(res.status).toBe(200);
+    await drain();
+
+    // The whole bug: this used to be zero calls and one logged
+    // "Original file not found" per photo.
+    expect(imageProcessor.ensureThumbnail).toHaveBeenCalledTimes(1);
+  });
+
+  it('nulls thumbnail_path so the valid-thumbnail short-circuit cannot skip the rebuild', async () => {
+    const eventId = await seedEvent();
+    await seedPhoto(eventId, {
+      source_origin: 'external',
+      external_relpath: 'shot.jpg',
+      thumbnail_path: 'thumbnails/still-on-disk.jpg',
+    });
+
+    await request(app).post('/admin/thumbnails/regenerate').send({});
+    await drain();
+
+    // Without this the endpoint is a no-op whenever the OLD thumbnail is still
+    // readable — which is the normal case after a settings change, and exactly
+    // when the admin pressed the button.
+    const [photoArg] = imageProcessor.ensureThumbnail.mock.calls[0];
+    expect(photoArg.thumbnail_path).toBeNull();
+    expect(photoArg.source_origin).toBe('external');
+    // Carried through so ensureThumbnail can resolve off the mount rather than
+    // under events/active.
+    expect(photoArg.external_relpath).toBe('shot.jpg');
+  });
+
+  it('leaves videos alone rather than handing a container file to Sharp', async () => {
+    const eventId = await seedEvent();
+    await seedPhoto(eventId, { source_origin: 'managed', media_type: 'video', filename: 'clip.mp4' });
+    await seedPhoto(eventId, { source_origin: 'managed', filename: 'still.jpg' });
+
+    const res = await request(app).post('/admin/thumbnails/regenerate').send({});
+    await drain();
+
+    expect(res.body.count).toBe(1);
+    expect(imageProcessor.ensureThumbnail).toHaveBeenCalledTimes(1);
+    expect(imageProcessor.ensureThumbnail.mock.calls[0][0].filename).toBe('still.jpg');
+  });
+
+  /**
+   * On S3, ensureThumbnail downloads the source to a randomly-named temp file,
+   * and for non-RAW input withProcessableImage passes no outputBasename — so
+   * generateThumbnail derives the key from that random name and it differs on
+   * every run. Nulling thumbnail_path hides the old key from everything that
+   * would otherwise clean it up, so each regeneration would strand a full
+   * thumbnail in the bucket, once per photo per run.
+   */
+  describe('superseded canonical renditions', () => {
+    it('removes the old thumbnail when the key moved', async () => {
+      const eventId = await seedEvent();
+      await seedPhoto(eventId, {
+        source_origin: 'managed',
+        thumbnail_path: 'thumbnails/thumb_OLDRANDOM_shot.jpg',
+      });
+      imageProcessor.ensureThumbnail.mockResolvedValueOnce('thumbnails/thumb_NEWRANDOM_shot.jpg');
+
+      await request(app).post('/admin/thumbnails/regenerate').send({});
+      await drain();
+
+      expect(storage.delete).toHaveBeenCalledWith('thumbnails/thumb_OLDRANDOM_shot.jpg');
+    });
+
+    it('does NOT delete when the key is unchanged — that is the new file', async () => {
+      const eventId = await seedEvent();
+      await seedPhoto(eventId, {
+        source_origin: 'managed',
+        thumbnail_path: 'thumbnails/thumb_stable.jpg',
+      });
+      // Local storage resolves to a stable path, so the key is identical.
+      imageProcessor.ensureThumbnail.mockResolvedValueOnce('thumbnails/thumb_stable.jpg');
+
+      await request(app).post('/admin/thumbnails/regenerate').send({});
+      await drain();
+
+      expect(storage.delete).not.toHaveBeenCalled();
+    });
+
+    it.each([
+      ['a Windows-style legacy path', 'thumbnails\\thumb_ext1_shot.jpg'],
+      ['a leading ./', './thumbnails/thumb_ext1_shot.jpg'],
+      ['a doubled separator', 'thumbnails//thumb_ext1_shot.jpg'],
+    ])('does not delete the file it just wrote when the old path is %s', async (_name, stored) => {
+      const eventId = await seedEvent();
+      await seedPhoto(eventId, { source_origin: 'managed', thumbnail_path: stored });
+      // Both storage backends fold these to the same key, so this is the SAME
+      // object — deleting it would remove the freshly generated thumbnail and
+      // leave the row pointing at nothing.
+      imageProcessor.ensureThumbnail.mockResolvedValueOnce('thumbnails/thumb_ext1_shot.jpg');
+
+      await request(app).post('/admin/thumbnails/regenerate').send({});
+      await drain();
+
+      expect(storage.delete).not.toHaveBeenCalled();
+    });
+
+    it('counts the photo as regenerated even if the old object cannot be removed', async () => {
+      const eventId = await seedEvent();
+      await seedPhoto(eventId, {
+        source_origin: 'managed',
+        thumbnail_path: 'thumbnails/thumb_OLD.jpg',
+      });
+      imageProcessor.ensureThumbnail.mockResolvedValueOnce('thumbnails/thumb_NEW.jpg');
+      storage.delete.mockRejectedValueOnce(new Error('bucket said no'));
+
+      await request(app).post('/admin/thumbnails/regenerate').send({});
+      await drain();
+
+      // Losing the old object is untidy; the regeneration itself succeeded.
+      expect(imageProcessor.ensureThumbnail).toHaveBeenCalledTimes(1);
+    });
+  });
+
+  it('scopes to one event when asked', async () => {
+    const a = await seedEvent();
+    await seedPhoto(a, { source_origin: 'external', external_relpath: 'a.jpg' });
+    const [b] = await db('events').insert({
+      slug: 'other', event_type: 'wedding', event_name: 'other', event_date: '2026-01-01',
+      host_email: 'h@example.com', admin_email: 'a@example.com', password_hash: 'x',
+      share_link: 'other-share', expires_at: new Date().toISOString(),
+    }).returning('id');
+    await seedPhoto(typeof b === 'object' ? b.id : b, { source_origin: 'managed' });
+
+    const res = await request(app).post('/admin/thumbnails/regenerate').send({ eventId: a });
+    await drain();
+
+    expect(res.body.count).toBe(1);
+    expect(imageProcessor.ensureThumbnail).toHaveBeenCalledTimes(1);
+  });
+});

backend/__tests__/integration/backup-s3.test.js

@@ -0,0 +1,524 @@
+const { describe, it, expect, beforeAll, afterAll, beforeEach, afterEach } = require('@jest/globals');
+const { S3Client, CreateBucketCommand, DeleteBucketCommand, ListObjectsV2Command, DeleteObjectsCommand } = require('@aws-sdk/client-s3');
+const path = require('path');
+const fs = require('fs').promises;
+const crypto = require('crypto');
+
+// Load services
+const backupService = require('../../src/services/backupService');
+const S3StorageAdapter = require('../../src/services/storage/s3Storage');
+const { db, initializeDatabase: initDb } = require('../../src/database/db');
+const logger = require('../../src/utils/logger');
+
+// Test configuration
+// Defaults match the dev MinIO container in docker-compose.dev.yml (port 7104).
+// Override via TEST_S3_ENDPOINT / TEST_S3_ACCESS_KEY / TEST_S3_SECRET_KEY when running
+// against a different S3 endpoint (CI, hosted MinIO, real AWS, etc.).
+const TEST_CONFIG = {
+  endpoint: process.env.TEST_S3_ENDPOINT || 'http://localhost:7104',
+  accessKeyId: process.env.TEST_S3_ACCESS_KEY || 'minioadmin',
+  secretAccessKey: process.env.TEST_S3_SECRET_KEY || 'minioadmin',
+  bucket: 'test-backup-bucket-' + Date.now(),
+  region: 'us-east-1'
+};
+
+describe('S3 Backup Integration Tests', () => {
+  let s3Client;
+  let testStoragePath;
+  let originalEnv;
+
+  beforeAll(async () => {
+    // Skip if no S3 endpoint configured
+    if (process.env.SKIP_S3_TESTS === 'true') {
+      console.log('Skipping S3 integration tests (SKIP_S3_TESTS=true)');
+      return;
+    }
+
+    // Save original environment
+    originalEnv = { ...process.env };
+
+    // Initialize S3 client for test setup
+    s3Client = new S3Client({
+      endpoint: TEST_CONFIG.endpoint,
+      region: TEST_CONFIG.region,
+      credentials: {
+        accessKeyId: TEST_CONFIG.accessKeyId,
+        secretAccessKey: TEST_CONFIG.secretAccessKey
+      },
+      forcePathStyle: true
+    });
+
+    // Create test bucket
+    try {
+      await s3Client.send(new CreateBucketCommand({ Bucket: TEST_CONFIG.bucket }));
+      console.log(`Created test bucket: ${TEST_CONFIG.bucket}`);
+    } catch (error) {
+      if (error.name !== 'BucketAlreadyOwnedByYou') {
+        console.error('Failed to create test bucket:', error);
+        throw error;
+      }
+    }
+
+    // Schema is expected to already be applied by `npm run migrate` against
+    // the dev database. db.migrate.latest() can't be used here because
+    // PicPeak's custom run-migrations.js tracks state in the `migrations`
+    // table (not knex's `knex_migrations`), so knex would try to re-apply
+    // every migration and crash on duplicate-table errors.
+    const ok = await db.schema.hasTable('events')
+      && await db.schema.hasTable('app_settings')
+      && await db.schema.hasTable('backup_runs');
+    if (!ok) {
+      throw new Error('Required tables missing — run `npm run migrate` against the dev DB first.');
+    }
+
+    // Create test storage directory
+    testStoragePath = path.join(__dirname, '../fixtures/test-storage');
+    await fs.mkdir(testStoragePath, { recursive: true });
+    process.env.STORAGE_PATH = testStoragePath;
+
+    // Set up test data
+    await setupTestData();
+
+    // Mock logger to reduce noise
+    if (process.env.UNMOCK_LOGGER !== 'true') {
+      logger.info = jest.fn();
+      logger.debug = jest.fn();
+      logger.warn = jest.fn();
+      logger.error = jest.fn();
+    }
+  });
+
+  afterAll(async () => {
+    if (process.env.SKIP_S3_TESTS === 'true') return;
+
+    try {
+      // Clean up S3 bucket
+      await cleanupS3Bucket();
+      await s3Client.send(new DeleteBucketCommand({ Bucket: TEST_CONFIG.bucket }));
+      console.log(`Deleted test bucket: ${TEST_CONFIG.bucket}`);
+    } catch (error) {
+      console.error('Failed to cleanup S3 bucket:', error);
+    }
+
+    // Clean up test storage
+    await fs.rm(testStoragePath, { recursive: true, force: true });
+
+    // Restore environment
+    process.env = originalEnv;
+
+    // Close database
+    await db.destroy();
+  });
+
+  beforeEach(async () => {
+    if (process.env.SKIP_S3_TESTS === 'true') {
+      return;
+    }
+
+    // Clean backup tables
+    await db('backup_runs').del();
+    await db('backup_file_states').del();
+    await db('database_backup_runs').del();
+
+    // Configure S3 backup settings
+    await configureS3Backup();
+  });
+
+  afterEach(async () => {
+    if (process.env.SKIP_S3_TESTS === 'true') return;
+
+    // Clean up S3 objects created during test
+    await cleanupS3Bucket();
+  });
+
+  describe('S3 Connection and Configuration', () => {
+    it('should successfully connect to S3-compatible storage', async () => {
+      if (process.env.SKIP_S3_TESTS === 'true') return;
+
+      const s3Adapter = new S3StorageAdapter({
+        ...TEST_CONFIG,
+        bucket: TEST_CONFIG.bucket,
+        forcePathStyle: true,
+        sslEnabled: false
+      });
+
+      const connected = await s3Adapter.testConnection();
+      expect(connected).toBe(true);
+    });
+
+    it('should validate S3 configuration before backup', async () => {
+      if (process.env.SKIP_S3_TESTS === 'true') return;
+
+      // Remove required configuration
+      await db('app_settings')
+        .where('setting_key', 'backup_s3_secret_key')
+        .del();
+
+      await backupService.runBackup();
+
+      const lastRun = await db('backup_runs')
+        .orderBy('started_at', 'desc')
+        .first();
+
+      expect(lastRun.status).toBe('failed');
+      expect(lastRun.error_message).toContain('S3 backup configuration incomplete');
+    });
+  });
+
+  describe('Full S3 Backup Process', () => {
+    it('should perform complete S3 backup with all file types', async () => {
+      if (process.env.SKIP_S3_TESTS === 'true') return;
+
+      // Run backup
+      await backupService.runBackup();
+
+      // Verify backup run completed
+      const backupRun = await db('backup_runs')
+        .orderBy('started_at', 'desc')
+        .first();
+
+      expect(backupRun.status).toBe('completed');
+      // pg driver returns bigint columns as strings; coerce for the size assertion.
+      expect(Number(backupRun.files_backed_up)).toBeGreaterThan(0);
+      expect(Number(backupRun.total_size_bytes)).toBeGreaterThan(0);
+
+      // Verify files in S3
+      const s3Objects = await listS3Objects();
+      expect(s3Objects.length).toBeGreaterThan(0);
+
+      // Check for expected file types
+      const hasPhotos = s3Objects.some(obj => obj.Key.includes('events/active'));
+      const hasThumbnails = s3Objects.some(obj => obj.Key.includes('thumbnails'));
+      const hasManifest = s3Objects.some(obj => obj.Key.includes('backup-manifest'));
+      const hasSummary = s3Objects.some(obj => obj.Key.includes('backup-summary.json'));
+
+      expect(hasPhotos).toBe(true);
+      expect(hasThumbnails).toBe(true);
+      expect(hasManifest).toBe(true);
+      expect(hasSummary).toBe(true);
+    });
+
+    it('should handle large file uploads with multipart', async () => {
+      if (process.env.SKIP_S3_TESTS === 'true') return;
+
+      // Create a large test file (15MB)
+      const largeFilePath = path.join(testStoragePath, 'events/active/large-photo.jpg');
+      const largeFileSize = 15 * 1024 * 1024; // 15MB
+      const largeFileContent = Buffer.alloc(largeFileSize, 'x');
+      await fs.writeFile(largeFilePath, largeFileContent);
+
+      // Run backup
+      await backupService.runBackup();
+
+      // Verify large file was uploaded
+      const s3Objects = await listS3Objects();
+      const largeFileUploaded = s3Objects.some(obj => 
+        obj.Key.includes('large-photo.jpg') && obj.Size === largeFileSize
+      );
+
+      expect(largeFileUploaded).toBe(true);
+    });
+
+    it('should include database backup when available', async () => {
+      if (process.env.SKIP_S3_TESTS === 'true') return;
+
+      // Create a mock database backup
+      const dbBackupPath = path.join(testStoragePath, 'backups/db-backup.sql');
+      await fs.mkdir(path.dirname(dbBackupPath), { recursive: true });
+      await fs.writeFile(dbBackupPath, 'CREATE TABLE test (id INT);');
+
+      // Record database backup
+      await db('database_backup_runs').insert({
+        started_at: new Date(),
+        completed_at: new Date(),
+        status: 'completed',
+        backup_type: 'sqlite',
+        file_path: dbBackupPath,
+        file_size_bytes: 100,
+        checksum: 'test123',
+        statistics: JSON.stringify({ tables: {} }),
+        table_checksums: JSON.stringify({})
+      });
+
+      // Configure to include database
+      await db('app_settings')
+        .where('setting_key', 'backup_include_database')
+        .update({ setting_value: 'true' });
+
+      // Run backup
+      await backupService.runBackup();
+
+      // Verify database backup in S3
+      const s3Objects = await listS3Objects();
+      const hasDbBackup = s3Objects.some(obj => obj.Key.includes('database/db-backup.sql'));
+      expect(hasDbBackup).toBe(true);
+    });
+  });
+
+  describe('Incremental Backup', () => {
+    it('should only upload changed files in incremental backup', async () => {
+      if (process.env.SKIP_S3_TESTS === 'true') return;
+
+      // First backup - full
+      await backupService.runBackup();
+
+      const firstRun = await db('backup_runs')
+        .orderBy('started_at', 'desc')
+        .first();
+
+      const firstObjectCount = (await listS3Objects()).length;
+
+      // Wait a moment to ensure different timestamps
+      await new Promise(resolve => setTimeout(resolve, 100));
+
+      // Modify one file
+      const modifiedFile = path.join(testStoragePath, 'events/active/event1/photo1.jpg');
+      await fs.writeFile(modifiedFile, 'modified content');
+
+      // Second backup - incremental
+      await backupService.runBackup();
+
+      const secondRun = await db('backup_runs')
+        .orderBy('started_at', 'desc')
+        .first();
+
+      expect(secondRun.id).not.toBe(firstRun.id);
+      expect(Number(secondRun.files_backed_up)).toBe(1); // Only modified file
+
+      // Check manifest indicates incremental. The current manifest schema
+      // groups counts under `incremental.changes.*` (added/modified/deleted/
+      // unchanged + size_difference) — see backupManifest.generateIncrementalManifest.
+      if (secondRun.manifest_path) {
+        const manifest = await backupService.getBackupManifest(secondRun.id);
+        expect(manifest.manifest.incremental).toBeDefined();
+        expect(manifest.manifest.incremental.changes).toBeDefined();
+        expect(manifest.manifest.incremental.changes.modified_files_count).toBe(1);
+      }
+    });
+
+    it('should track file states across backups', async () => {
+      if (process.env.SKIP_S3_TESTS === 'true') return;
+
+      await backupService.runBackup();
+
+      // Check file states are recorded
+      const fileStates = await db('backup_file_states').select('*');
+      expect(fileStates.length).toBeGreaterThan(0);
+
+      // Verify checksums are stored
+      const hasChecksums = fileStates.every(state => state.checksum !== null);
+      expect(hasChecksums).toBe(true);
+    });
+  });
+
+  describe('S3 Manifest Storage', () => {
+    it('should upload manifest to S3 and retrieve it', async () => {
+      if (process.env.SKIP_S3_TESTS === 'true') return;
+
+      // Configure YAML manifest format
+      await db('app_settings')
+        .where('setting_key', 'backup_manifest_format')
+        .update({ setting_value: '"yaml"' });
+
+      await backupService.runBackup();
+
+      const backupRun = await db('backup_runs')
+        .orderBy('started_at', 'desc')
+        .first();
+
+      expect(backupRun.manifest_path).toMatch(/^s3:\/\//);
+
+      // Retrieve manifest
+      const { manifest, summary } = await backupService.getBackupManifest(backupRun.id);
+
+      expect(manifest).toBeDefined();
+      expect(manifest.backup.id).toBeDefined();
+      expect(summary).toContain('BACKUP MANIFEST SUMMARY');
+    });
+
+    it('should validate manifest integrity', async () => {
+      if (process.env.SKIP_S3_TESTS === 'true') return;
+
+      await backupService.runBackup();
+
+      const backupRun = await db('backup_runs')
+        .orderBy('started_at', 'desc')
+        .first();
+
+      const validationResult = await backupService.validateBackupManifest(backupRun.manifest_path);
+
+      expect(validationResult.valid).toBe(true);
+      expect(validationResult.manifest).toBeDefined();
+    });
+  });
+
+  describe('Error Recovery', () => {
+    it('should handle S3 connection failures gracefully', async () => {
+      if (process.env.SKIP_S3_TESTS === 'true') return;
+
+      // Configure with invalid endpoint
+      await db('app_settings')
+        .where('setting_key', 'backup_s3_endpoint')
+        .update({ setting_value: '"http://invalid-endpoint:9999"' });
+
+      await backupService.runBackup();
+
+      const backupRun = await db('backup_runs')
+        .orderBy('started_at', 'desc')
+        .first();
+
+      expect(backupRun.status).toBe('failed');
+      expect(backupRun.error_message).toBeDefined();
+    });
+
+    it('should continue backup despite individual file failures', async () => {
+      if (process.env.SKIP_S3_TESTS === 'true') return;
+
+      // Create a file that will be deleted during backup
+      const tempFile = path.join(testStoragePath, 'events/active/temp.jpg');
+      await fs.writeFile(tempFile, 'temporary');
+
+      // Mock file deletion during backup
+      const originalUpload = S3StorageAdapter.prototype.upload;
+      let callCount = 0;
+      S3StorageAdapter.prototype.upload = jest.fn(async function(localPath, s3Key, options) {
+        callCount++;
+        if (callCount === 2) {
+          // Delete the temp file to cause an error
+          await fs.unlink(tempFile).catch(() => {});
+        }
+        return originalUpload.call(this, localPath, s3Key, options);
+      });
+
+      await backupService.runBackup();
+
+      const backupRun = await db('backup_runs')
+        .orderBy('started_at', 'desc')
+        .first();
+
+      // Should complete despite one file error
+      expect(backupRun.status).toBe('completed');
+      expect(backupRun.files_backed_up).toBeGreaterThan(0);
+
+      // Restore original method
+      S3StorageAdapter.prototype.upload = originalUpload;
+    });
+
+    it('should retry failed uploads with exponential backoff', async () => {
+      if (process.env.SKIP_S3_TESTS === 'true') return;
+
+      // Mock S3 upload to fail twice then succeed
+      const originalUpload = S3StorageAdapter.prototype.upload;
+      let attemptCount = 0;
+      S3StorageAdapter.prototype.upload = jest.fn(async function(localPath, s3Key, options) {
+        attemptCount++;
+        if (attemptCount <= 2) {
+          const error = new Error('Network timeout');
+          error.code = 'ETIMEDOUT';
+          throw error;
+        }
+        return originalUpload.call(this, localPath, s3Key, options);
+      });
+
+      await backupService.runBackup();
+
+      const backupRun = await db('backup_runs')
+        .orderBy('started_at', 'desc')
+        .first();
+
+      // Should succeed after retries
+      expect(backupRun.status).toBe('completed');
+      expect(attemptCount).toBeGreaterThan(2);
+
+      // Restore original method
+      S3StorageAdapter.prototype.upload = originalUpload;
+    });
+  });
+
+  // Helper functions
+
+  async function setupTestData() {
+    // Create test directory structure
+    const dirs = [
+      'events/active/event1',
+      'events/active/event2',
+      'events/archived',
+      'thumbnails',
+      'uploads'
+    ];
+
+    for (const dir of dirs) {
+      await fs.mkdir(path.join(testStoragePath, dir), { recursive: true });
+    }
+
+    // Create test files
+    const files = [
+      { path: 'events/active/event1/photo1.jpg', content: 'photo1 content' },
+      { path: 'events/active/event1/photo2.jpg', content: 'photo2 content' },
+      { path: 'events/active/event2/photo3.jpg', content: 'photo3 content' },
+      { path: 'events/archived/old-event.zip', content: 'archived content' },
+      { path: 'thumbnails/thumb1.jpg', content: 'thumbnail content' },
+      { path: 'uploads/logo.png', content: 'logo content' }
+    ];
+
+    for (const file of files) {
+      await fs.writeFile(
+        path.join(testStoragePath, file.path),
+        file.content
+      );
+    }
+  }
+
+  async function configureS3Backup() {
+    const settings = [
+      { setting_key: 'backup_enabled', setting_value: 'true' },
+      { setting_key: 'backup_destination_type', setting_value: '"s3"' },
+      { setting_key: 'backup_s3_bucket', setting_value: `"${TEST_CONFIG.bucket}"` },
+      { setting_key: 'backup_s3_region', setting_value: `"${TEST_CONFIG.region}"` },
+      { setting_key: 'backup_s3_endpoint', setting_value: `"${TEST_CONFIG.endpoint}"` },
+      { setting_key: 'backup_s3_access_key', setting_value: `"${TEST_CONFIG.accessKeyId}"` },
+      { setting_key: 'backup_s3_secret_key', setting_value: `"${TEST_CONFIG.secretAccessKey}"` },
+      { setting_key: 'backup_s3_force_path_style', setting_value: 'true' },
+      { setting_key: 'backup_s3_ssl_enabled', setting_value: 'false' },
+      { setting_key: 'backup_include_archived', setting_value: 'true' },
+      { setting_key: 'backup_incremental', setting_value: 'true' },
+      { setting_key: 'backup_max_file_size_mb', setting_value: '100' }
+    ];
+
+    // Schema drift: app_settings has no created_at column anymore and the
+    // unique constraint is on setting_key alone, not (setting_type, key).
+    for (const setting of settings) {
+      await db('app_settings')
+        .insert({
+          setting_type: 'backup',
+          ...setting,
+          updated_at: new Date(),
+        })
+        .onConflict('setting_key')
+        .merge();
+    }
+  }
+
+  async function listS3Objects() {
+    const response = await s3Client.send(new ListObjectsV2Command({
+      Bucket: TEST_CONFIG.bucket
+    }));
+    return response.Contents || [];
+  }
+
+  async function cleanupS3Bucket() {
+    try {
+      const objects = await listS3Objects();
+      if (objects.length > 0) {
+        await s3Client.send(new DeleteObjectsCommand({
+          Bucket: TEST_CONFIG.bucket,
+          Delete: {
+            Objects: objects.map(obj => ({ Key: obj.Key }))
+          }
+        }));
+      }
+    } catch (error) {
+      console.error('Failed to cleanup S3 objects:', error);
+    }
+  }
+});

backend/__tests__/integration/backupService.businessDocs.test.js

@@ -0,0 +1,88 @@
+/**
+ * Regression net for the business-docs coverage gap fixed in this PR.
+ *
+ * Prior to the fix, `getFilesToBackupInternal()` enumerated a fixed
+ * list of storage subdirectories (events/active, events/archived,
+ * thumbnails, previews, heroes, uploads) and silently omitted the
+ * entire `business-docs/` tree. That meant every CRM PDF + signature
+ * drawing — quotes, contracts (system-rendered + wet uploads),
+ * invoices, Storno, imported historical invoices, and the customer
+ * signature PNG/JPG drawn on the public signing page — fell outside
+ * the in-app scheduled backup, leaving every `*_path` column on
+ * `quotes` / `contracts` / `invoices` as a broken FK after restore.
+ *
+ * The fix is a single `scanDirectory(business-docs, ...)` call. This
+ * suite pins the contract so a future refactor of the walker cannot
+ * silently drop business-docs again.
+ */
+
+const fs = require('fs');
+const path = require('path');
+
+const { bootCrmDb } = require('./helpers/crmDb');
+
+describe('backupService — business-docs is in the backup walker', () => {
+  let cleanup;
+  let backupService;
+  let storagePath;
+
+  beforeAll(async () => {
+    ({ cleanup } = await bootCrmDb());
+    storagePath = process.env.STORAGE_PATH;
+    // Cold-require after bootCrmDb so backupService picks up the same
+    // db instance + STORAGE_PATH the test harness configured.
+    backupService = require('../../src/services/backupService');
+  }, 120000);
+
+  afterAll(async () => {
+    if (cleanup) await cleanup();
+  });
+
+  function seed(relPath, content = 'dummy bytes for backup test') {
+    const abs = path.join(storagePath, relPath);
+    fs.mkdirSync(path.dirname(abs), { recursive: true });
+    fs.writeFileSync(abs, content);
+  }
+
+  it('does not error when business-docs is absent', async () => {
+    // Fresh harness has no business-docs/ tree at all. The walker
+    // must short-circuit on ENOENT rather than throw — installs that
+    // never used CRM features have to keep backing up fine.
+    await expect(backupService.getFilesToBackup(false)).resolves.toEqual(expect.any(Array));
+  });
+
+  it('picks up every CRM-relevant business-docs subdirectory', async () => {
+    // Seed one file in each of the five subpaths the renderer + import
+    // routes write to. The signature path is the one most prone to be
+    // forgotten — it lives one level deeper than the others (per-
+    // contract subfolder, not per-year).
+    seed('business-docs/quote/2026/Q-001.pdf');
+    seed('business-docs/contract/2026/C-001.pdf');
+    seed('business-docs/contract/signatures/42/customer-1700000000000.png');
+    seed('business-docs/invoice/2026/INV-001.pdf');
+    seed('business-docs/invoice-imports/2026/scan.pdf');
+
+    const files = await backupService.getFilesToBackup(false);
+    const rels = files.map((f) => f.relativePath);
+
+    expect(rels).toEqual(expect.arrayContaining([
+      'business-docs/quote/2026/Q-001.pdf',
+      'business-docs/contract/2026/C-001.pdf',
+      'business-docs/contract/signatures/42/customer-1700000000000.png',
+      'business-docs/invoice/2026/INV-001.pdf',
+      'business-docs/invoice-imports/2026/scan.pdf',
+    ]));
+  });
+
+  it('walks newly-created business-docs files without needing a restart', async () => {
+    // The walker reads the filesystem live on every call; this guards
+    // against a future "cache the scan result at boot" optimisation
+    // that would miss freshly-written PDFs (which is exactly what
+    // happens during normal operation — every send writes a new file).
+    seed('business-docs/invoice/2027/INV-NEW.pdf');
+
+    const files = await backupService.getFilesToBackup(false);
+    const rels = files.map((f) => f.relativePath);
+    expect(rels).toContain('business-docs/invoice/2027/INV-NEW.pdf');
+  });
+});

backend/__tests__/integration/backupService.configurableWalker.test.js

@@ -0,0 +1,379 @@
+/**
+ * Pins the Stage-B refactor that lifted the file-backup walker's
+ * subdirectory list out of hard-coded JS into the `backup_paths`
+ * table seeded by migration 109.
+ *
+ * Scenarios:
+ *   1. Walker reads canonical seed → all 7 default subdirs walked
+ *   2. include_in_default=false on one row → that subdir is skipped
+ *   3. New row inserted at runtime → walker picks it up without restart
+ *   4. feature_flag gating → row only walked when the named app_settings
+ *      boolean is truthy (mirrors historical `includeArchived` behavior)
+ *   5. Empty table → walker falls back to LEGACY_BACKUP_PATHS (defense
+ *      in depth — never silently scans nothing)
+ *
+ * Why not stub `db('backup_paths')`: the whole point of Stage B is
+ * that the walker is now data-driven, so the test has to actually
+ * mutate the table and observe the walker's output change. Stubs
+ * would re-introduce the hard-coding the refactor is meant to remove.
+ */
+
+const fs = require('fs');
+const path = require('path');
+
+const { bootCrmDb } = require('./helpers/crmDb');
+
+jest.setTimeout(120000);
+
+describe('backupService — configurable walker (backup_paths)', () => {
+  let db;
+  let cleanup;
+  let storagePath;
+  let backupService;
+
+  beforeAll(async () => {
+    ({ db, cleanup } = await bootCrmDb());
+    storagePath = process.env.STORAGE_PATH;
+    backupService = require('../../src/services/backupService');
+  }, 120000);
+
+  afterAll(async () => {
+    if (cleanup) await cleanup();
+  });
+
+  function seedFile(relPath, content = 'dummy bytes') {
+    const abs = path.join(storagePath, relPath);
+    fs.mkdirSync(path.dirname(abs), { recursive: true });
+    fs.writeFileSync(abs, content);
+  }
+
+  beforeEach(async () => {
+    // Restore canonical seed before every test. Tests mutate this table
+    // freely; the next test starts from a known state.
+    await db('backup_paths').del();
+    const {
+      DEFAULT_PATHS,
+    } = require('../../migrations/core/109_add_backup_paths');
+    await db('backup_paths').insert(DEFAULT_PATHS.map((row) => ({
+      ...row,
+      created_at: new Date(),
+      updated_at: new Date(),
+    })));
+  });
+
+  it('migration 109 seeds the canonical 7 paths', async () => {
+    const rows = await db('backup_paths').orderBy('display_order', 'asc').select();
+    expect(rows.map((r) => r.path)).toEqual([
+      'events/active',
+      'events/archived',
+      'thumbnails',
+      'previews',
+      'heroes',
+      'uploads',
+      'business-docs',
+    ]);
+    // Only events/archived is gated by a feature flag.
+    expect(rows.filter((r) => r.feature_flag).map((r) => r.path)).toEqual([
+      'events/archived',
+    ]);
+  });
+
+  it('walks every default subdir when files are present', async () => {
+    seedFile('events/active/E1/a.jpg');
+    seedFile('thumbnails/E1/a.jpg');
+    seedFile('previews/E1/a.jpg');
+    seedFile('heroes/E1/hero.jpg');
+    seedFile('uploads/intake/x.bin');
+    seedFile('business-docs/quote/2026/Q-001.pdf');
+    // events/archived is gated — left out of this test; covered below.
+
+    const files = await backupService.getFilesToBackup({ backup_include_archived: true });
+    const rels = files.map((f) => f.relativePath);
+
+    expect(rels).toEqual(expect.arrayContaining([
+      'events/active/E1/a.jpg',
+      'thumbnails/E1/a.jpg',
+      'previews/E1/a.jpg',
+      'heroes/E1/hero.jpg',
+      'uploads/intake/x.bin',
+      'business-docs/quote/2026/Q-001.pdf',
+    ]));
+  });
+
+  it('skips a path when include_in_default is toggled off', async () => {
+    seedFile('thumbnails/E1/thumb.jpg');
+    seedFile('events/active/E1/photo.jpg');
+
+    await db('backup_paths').where('path', 'thumbnails').update({
+      include_in_default: false,
+    });
+
+    const files = await backupService.getFilesToBackup({ backup_include_archived: true });
+    const rels = files.map((f) => f.relativePath);
+
+    expect(rels).toContain('events/active/E1/photo.jpg');
+    expect(rels).not.toContain('thumbnails/E1/thumb.jpg');
+  });
+
+  it('picks up a new path inserted at runtime — no restart needed', async () => {
+    // Simulates a future feature shipping its own subdirectory and
+    // self-healing a `backup_paths` row at boot.
+    await db('backup_paths').insert({
+      path: 'plugin-store',
+      include_in_default: true,
+      feature_flag: null,
+      display_order: 200,
+      description: 'Hypothetical future feature payload',
+      created_at: new Date(),
+      updated_at: new Date(),
+    });
+    seedFile('plugin-store/cache/payload.bin');
+
+    const files = await backupService.getFilesToBackup({ backup_include_archived: true });
+    const rels = files.map((f) => f.relativePath);
+
+    expect(rels).toContain('plugin-store/cache/payload.bin');
+  });
+
+  it('respects feature_flag gating (events/archived ⇄ backup_include_archived)', async () => {
+    seedFile('events/active/E1/active.jpg');
+    seedFile('events/archived/E2/archived.jpg');
+
+    // backup_include_archived=false → archived/ is skipped.
+    const filesOff = await backupService.getFilesToBackup({ backup_include_archived: false });
+    const relsOff = filesOff.map((f) => f.relativePath);
+    expect(relsOff).toContain('events/active/E1/active.jpg');
+    expect(relsOff).not.toContain('events/archived/E2/archived.jpg');
+
+    // backup_include_archived=true → archived/ is included.
+    const filesOn = await backupService.getFilesToBackup({ backup_include_archived: true });
+    const relsOn = filesOn.map((f) => f.relativePath);
+    expect(relsOn).toContain('events/archived/E2/archived.jpg');
+  });
+
+  it('falls back to LEGACY_BACKUP_PATHS when the table is empty', async () => {
+    // Defense in depth: even if seed-and-self-heal both failed, the
+    // walker must still cover the historical set so "Run Backup Now"
+    // cannot silently degrade to no-op.
+    await db('backup_paths').del();
+    seedFile('events/active/E1/photo.jpg');
+    seedFile('business-docs/quote/2026/Q-002.pdf');
+
+    const files = await backupService.getFilesToBackup({ backup_include_archived: true });
+    const rels = files.map((f) => f.relativePath);
+
+    expect(rels).toContain('events/active/E1/photo.jpg');
+    expect(rels).toContain('business-docs/quote/2026/Q-002.pdf');
+  });
+
+  it('legacy boolean call signature still works (backward compat)', async () => {
+    // Existing call sites (and the businessDocs regression test) pass
+    // a boolean for `includeArchived`. Refactor must not break them.
+    seedFile('events/archived/E3/legacy.jpg');
+
+    const filesOff = await backupService.getFilesToBackup(false);
+    expect(filesOff.map((f) => f.relativePath)).not.toContain('events/archived/E3/legacy.jpg');
+
+    const filesOn = await backupService.getFilesToBackup(true);
+    expect(filesOn.map((f) => f.relativePath)).toContain('events/archived/E3/legacy.jpg');
+  });
+
+  // Issue #871 — the "What to Backup" checkboxes were stored but never read.
+  describe('UI opt-out toggles (issue #871)', () => {
+    it('unchecking Thumbnails excludes thumbnails/', async () => {
+      seedFile('thumbnails/E1/thumb.jpg');
+      seedFile('events/active/E1/photo.jpg');
+
+      const files = await backupService.getFilesToBackup({
+        backup_include_thumbnails: false,
+      });
+      const rels = files.map((f) => f.relativePath);
+
+      expect(rels).toContain('events/active/E1/photo.jpg');
+      expect(rels).not.toContain('thumbnails/E1/thumb.jpg');
+    });
+
+    it('unchecking Photos excludes events/active', async () => {
+      seedFile('thumbnails/E1/thumb.jpg');
+      seedFile('events/active/E1/photo.jpg');
+
+      const files = await backupService.getFilesToBackup({
+        backup_include_photos: false,
+      });
+      const rels = files.map((f) => f.relativePath);
+
+      expect(rels).toContain('thumbnails/E1/thumb.jpg');
+      expect(rels).not.toContain('events/active/E1/photo.jpg');
+    });
+
+    it('defaults to including everything when the keys were never saved', async () => {
+      seedFile('thumbnails/E1/thumb.jpg');
+      seedFile('events/active/E1/photo.jpg');
+
+      const files = await backupService.getFilesToBackup({});
+      const rels = files.map((f) => f.relativePath);
+
+      expect(rels).toContain('thumbnails/E1/thumb.jpg');
+      expect(rels).toContain('events/active/E1/photo.jpg');
+    });
+
+    it("accepts the UI's plural backup_include_archives for the archived gate", async () => {
+      seedFile('events/archived/E4/archived.jpg');
+
+      const files = await backupService.getFilesToBackup({
+        backup_include_archives: true,
+      });
+      expect(files.map((f) => f.relativePath)).toContain('events/archived/E4/archived.jpg');
+    });
+
+    it('the UI plural key beats the migration-seeded singular key', async () => {
+      // Migration seeds backup_include_archived=true on every install; the
+      // form only ever writes the plural key, so unchecking Archives must
+      // win over the stale seeded value.
+      seedFile('events/archived/E5/archived.jpg');
+
+      const files = await backupService.getFilesToBackup({
+        backup_include_archived: true,   // seeded default
+        backup_include_archives: false,  // what the admin actually chose
+      });
+      expect(files.map((f) => f.relativePath)).not.toContain('events/archived/E5/archived.jpg');
+    });
+
+    it('rsync gets the de-selected paths and noise filters as --exclude args', async () => {
+      const excluded = await backupService.resolveExcludedBackupPaths({
+        backup_include_thumbnails: false,
+        backup_include_archives: false,
+      });
+      expect(excluded.map((r) => r.path)).toEqual(
+        expect.arrayContaining(['thumbnails', 'events/archived'])
+      );
+
+      const args = backupService.buildRsyncArgs(
+        { backup_rsync_host: 'backup.example.com', backup_rsync_path: '/srv/backups' },
+        excluded.map((r) => `/${r.path}/`)
+      );
+      const excludes = args
+        .map((a, i) => (a === '--exclude' ? args[i + 1] : null))
+        .filter(Boolean);
+      expect(excludes).toEqual(expect.arrayContaining([
+        '.nfs*',
+        '/thumbnails/',
+        '/events/archived/',
+      ]));
+    });
+
+    it('rows toggled off via include_in_default also become rsync excludes', async () => {
+      // The enabled-only loader hides these rows from the walker, but rsync
+      // syncs the whole storage root, so they must still appear as excludes.
+      await db('backup_paths').where('path', 'previews').update({
+        include_in_default: false,
+      });
+
+      const excluded = await backupService.resolveExcludedBackupPaths({});
+      expect(excluded.map((r) => r.path)).toContain('previews');
+    });
+  });
+
+  // Issue #871 — .nfs* silly-rename artifacts were uploaded to S3.
+  it('never backs up filesystem noise (.nfs*, .DS_Store)', async () => {
+    seedFile('thumbnails/E1/.nfs000000000000006600000008');
+    seedFile('events/active/E1/.DS_Store');
+    seedFile('events/active/E1/photo.jpg');
+
+    const files = await backupService.getFilesToBackup({});
+    const rels = files.map((f) => f.relativePath);
+
+    expect(rels).toContain('events/active/E1/photo.jpg');
+    expect(rels.some((r) => r.includes('.nfs'))).toBe(false);
+    expect(rels.some((r) => r.includes('.DS_Store'))).toBe(false);
+  });
+
+  it('the walker honors backup_exclude_patterns (previously rsync-only)', async () => {
+    seedFile('events/active/E1/photo.jpg');
+    seedFile('events/active/E1/scratch.tmp');
+
+    const files = await backupService.getFilesToBackup({
+      backup_exclude_patterns: ['*.tmp'],
+    });
+    const rels = files.map((f) => f.relativePath);
+
+    expect(rels).toContain('events/active/E1/photo.jpg');
+    expect(rels).not.toContain('events/active/E1/scratch.tmp');
+  });
+
+  it('glob patterns are literal outside the star (.nfs* must not eat anfs-…)', async () => {
+    seedFile('events/active/E1/anfs-photo.jpg');
+    seedFile('events/active/E1/notes-tmp');
+
+    const files = await backupService.getFilesToBackup({
+      backup_exclude_patterns: ['*.tmp'],
+    });
+    const rels = files.map((f) => f.relativePath);
+
+    // '.nfs*' used to compile to /^.nfs.*$/ whose dot matched any char;
+    // '*.tmp' used to compile to /^.*.tmp$/ which also matched 'notes-tmp'.
+    expect(rels).toContain('events/active/E1/anfs-photo.jpg');
+    expect(rels).toContain('events/active/E1/notes-tmp');
+  });
+
+  // Issue #871 — weekly schedules silently ran daily, and the dashboard's
+  // "next backup" was a hardcoded "tomorrow 02:00".
+  describe('schedule resolution + next run (issue #871)', () => {
+    it('a named label beats the stray default cron the UI used to send', () => {
+      expect(backupService.resolveScheduleCron({
+        backup_schedule: 'weekly',
+        backup_schedule_cron: '0 3 * * *', // old UI default, sent unconditionally
+      })).toBe('0 3 * * 0');
+    });
+
+    it('custom schedules use the cron field', () => {
+      expect(backupService.resolveScheduleCron({
+        backup_schedule: 'custom',
+        backup_schedule_cron: '15 5 * * 2',
+      })).toBe('15 5 * * 2');
+    });
+
+    it('falls back to the default daily cron', () => {
+      expect(backupService.resolveScheduleCron({})).toBe('0 2 * * *');
+    });
+
+    it('getNextScheduledRun is null when backups are disabled', () => {
+      expect(backupService.getNextScheduledRun(null)).toBeNull();
+      expect(backupService.getNextScheduledRun({ backup_enabled: false })).toBeNull();
+    });
+
+    it('getNextScheduledRun returns the real next weekly fire time', () => {
+      const iso = backupService.getNextScheduledRun({
+        backup_enabled: true,
+        backup_schedule: 'weekly',
+        backup_schedule_cron: '0 3 * * *',
+      });
+      const next = new Date(iso);
+      expect(Number.isNaN(next.getTime())).toBe(false);
+      expect(next.getTime()).toBeGreaterThan(Date.now());
+      expect(next.getDay()).toBe(0);   // Sunday
+      expect(next.getHours()).toBe(3); // 03:00
+    });
+  });
+
+  // Issue #871 — "Backup Size: 167.6 TB": file_size_bytes is a bigInteger
+  // column, node-postgres returns int8 as a string, and the S3 path did
+  // `backedUpSize += size` — string concatenation.
+  it('getDatabaseBackupInfo coerces file_size_bytes to a number', async () => {
+    await db('database_backup_runs').del();
+    await db('database_backup_runs').insert({
+      backup_type: 'full',
+      status: 'completed',
+      file_path: '/backups/db/dump.sql.gz',
+      // Simulate the PG int8-as-string driver behaviour (sqlite stores
+      // whatever it is handed, so the string round-trips).
+      file_size_bytes: '421988',
+      started_at: new Date().toISOString(),
+      completed_at: new Date().toISOString(),
+    });
+
+    const info = await backupService.getDatabaseBackupInfo();
+    expect(typeof info.size).toBe('number');
+    expect(info.size).toBe(421988);
+  });
+});

backend/__tests__/integration/backupService.inlineDbDump.test.js

@@ -0,0 +1,188 @@
+/**
+ * Pins the inline-DB-dump + fail-loud guard added to `runBackupInternal`.
+ *
+ * The previous behaviour was: file-backup looked up an existing dump via
+ * `getDatabaseBackupInfo()` and silently shipped a files-only manifest
+ * when none was found. Admins clicking "Run Backup Now" got an apparent
+ * success that omitted every customer / quote / invoice / contract row —
+ * the data-loss footgun that this commit closes.
+ *
+ * Five scenarios under test:
+ *   1. Default (inline dump enabled), dump succeeds → backup proceeds
+ *   2. Default, dump throws → run aborts, backup_runs row marked failed
+ *   3. Opt-out + recent DB dump available → backup proceeds
+ *   4. Opt-out + no DB dump available → fail loud
+ *   5. Opt-out + DB dump file is 0 bytes on disk → fail loud
+ *
+ * Mocking strategy: the underlying `databaseBackupService.backup()` and
+ * the local-destination writer are stubbed so the test exercises just
+ * the new guard logic without depending on `pg_dump` / `sqlite3` CLI
+ * binaries being available in the test environment.
+ */
+
+const fs = require('fs');
+const path = require('path');
+
+const { bootCrmDb } = require('./helpers/crmDb');
+
+// Set up mocks BEFORE bootCrmDb so backupService picks them up at require time.
+const mockBackupFn = jest.fn();
+jest.mock('../../src/services/databaseBackup', () => ({
+  databaseBackupService: { backup: mockBackupFn },
+  startScheduledBackups: jest.fn(),
+  stopScheduledBackups: jest.fn(),
+  DatabaseBackupService: class {},
+}));
+
+jest.setTimeout(120000);
+
+describe('backupService — inline DB dump + fail-loud guard', () => {
+  let db;
+  let cleanup;
+  let storagePath;
+  let backupService;
+  let dumpFileAbs;
+
+  beforeAll(async () => {
+    ({ db, cleanup } = await bootCrmDb());
+    storagePath = process.env.STORAGE_PATH;
+    backupService = require('../../src/services/backupService');
+
+    // Seed backup destination settings so the run can proceed past the
+    // "destination not configured" guard.
+    const dest = path.join(storagePath, 'backups');
+    fs.mkdirSync(dest, { recursive: true });
+    // getBackupConfigInternal filters by setting_type='backup', so the
+    // tests have to seed with that type or the resolver returns
+    // `{ ... }` with the keys missing — runBackup then sees
+    // `backup_destination_type === undefined` and bails before our
+    // new guard runs.
+    await db('app_settings').insert([
+      { setting_key: 'backup_destination_type',    setting_value: JSON.stringify('local'), setting_type: 'backup' },
+      { setting_key: 'backup_destination_path',    setting_value: JSON.stringify(dest),    setting_type: 'backup' },
+      { setting_key: 'backup_enabled',             setting_value: JSON.stringify(true),    setting_type: 'backup' },
+      { setting_key: 'backup_email_on_failure',    setting_value: JSON.stringify(false),   setting_type: 'backup' },
+    ]).onConflict('setting_key').merge();
+
+    // Pre-create a dump file that getDatabaseBackupInfo can resolve to.
+    // Reused/mutated per-test via the database_backup_runs seed below.
+    dumpFileAbs = path.join(storagePath, 'backups', 'fake-dump.sql.gz');
+    fs.writeFileSync(dumpFileAbs, 'pretend this is a pg_dump'.repeat(100));
+
+    // Neutralise the file-scan step: we don't care which files would
+    // be backed up, just whether the run reaches that stage at all.
+    backupService.getFilesToBackup = jest.fn(async () => []);
+  }, 120000);
+
+  afterAll(async () => {
+    if (cleanup) await cleanup();
+  });
+
+  beforeEach(async () => {
+    mockBackupFn.mockReset();
+    // Default to "dump produced this file with this size" — the per-test
+    // setup overrides as needed.
+    mockBackupFn.mockResolvedValue({
+      success: true,
+      path: dumpFileAbs,
+      size: fs.statSync(dumpFileAbs).size,
+      duration: 1,
+      checksum: 'abc',
+    });
+
+    // Re-seed the database_backup_runs row that getDatabaseBackupInfo
+    // resolves against (its query is `status='completed'` + most recent).
+    await db('database_backup_runs').del();
+    await db('database_backup_runs').insert({
+      started_at: new Date(),
+      completed_at: new Date(),
+      status: 'completed',
+      backup_type: 'pg',
+      file_path: dumpFileAbs,
+      file_size_bytes: fs.statSync(dumpFileAbs).size,
+      destination_path: dumpFileAbs,
+    });
+  });
+
+  it('default behaviour: inline dump runs, then file backup proceeds', async () => {
+    // Inline-dump setting is unset (undefined) — default is ON.
+    await db('app_settings').where('setting_key', 'backup_database_inline_dump').del();
+
+    await backupService.runBackup(true);
+
+    expect(mockBackupFn).toHaveBeenCalledTimes(1);
+
+    const run = await db('backup_runs').orderBy('id', 'desc').first();
+    expect(run.status).toBe('completed');
+    expect(run.error_message).toBeNull();
+  });
+
+  it('aborts the run when the inline dump throws', async () => {
+    await db('app_settings').where('setting_key', 'backup_database_inline_dump').del();
+    mockBackupFn.mockRejectedValueOnce(new Error('pg_dump segfaulted'));
+
+    await backupService.runBackup(true);
+
+    const run = await db('backup_runs').orderBy('id', 'desc').first();
+    expect(run.status).toBe('failed');
+    expect(run.error_message).toMatch(/pg_dump segfaulted/);
+  });
+
+  it('opt-out: skips inline dump but proceeds when a recent dump exists', async () => {
+    await db('app_settings').insert({
+      setting_key: 'backup_database_inline_dump',
+      setting_value: JSON.stringify(false),
+      setting_type: 'backup',
+    }).onConflict('setting_key').merge();
+
+    await backupService.runBackup(true);
+
+    expect(mockBackupFn).not.toHaveBeenCalled();
+
+    const run = await db('backup_runs').orderBy('id', 'desc').first();
+    expect(run.status).toBe('completed');
+  });
+
+  it('opt-out + no recent dump: fails loud with a clear error', async () => {
+    await db('app_settings').insert({
+      setting_key: 'backup_database_inline_dump',
+      setting_value: JSON.stringify(false),
+      setting_type: 'backup',
+    }).onConflict('setting_key').merge();
+    // Wipe the dump row so getDatabaseBackupInfo returns backupFile=null.
+    await db('database_backup_runs').del();
+
+    await backupService.runBackup(true);
+
+    const run = await db('backup_runs').orderBy('id', 'desc').first();
+    expect(run.status).toBe('failed');
+    expect(run.error_message).toMatch(/No database backup available/);
+  });
+
+  it('opt-out + 0-byte dump file: fails loud', async () => {
+    await db('app_settings').insert({
+      setting_key: 'backup_database_inline_dump',
+      setting_value: JSON.stringify(false),
+      setting_type: 'backup',
+    }).onConflict('setting_key').merge();
+
+    const emptyDump = path.join(storagePath, 'backups', 'empty-dump.sql.gz');
+    fs.writeFileSync(emptyDump, '');
+    await db('database_backup_runs').del();
+    await db('database_backup_runs').insert({
+      started_at: new Date(),
+      completed_at: new Date(),
+      status: 'completed',
+      backup_type: 'pg',
+      file_path: emptyDump,
+      file_size_bytes: 0,
+      destination_path: emptyDump,
+    });
+
+    await backupService.runBackup(true);
+
+    const run = await db('backup_runs').orderBy('id', 'desc').first();
+    expect(run.status).toBe('failed');
+    expect(run.error_message).toMatch(/is empty/);
+  });
+});

backend/__tests__/integration/backupService.perPathStats.test.js

@@ -0,0 +1,180 @@
+/**
+ * Per-Stage-B-path tally — Tier 3 of tonight's backup hardening.
+ *
+ * Pins the new `computePerPathStats` logic that the Backup History
+ * "Content Backed Up" pane reads via `backup_runs.statistics.per_path`.
+ *
+ * Three scenarios:
+ *   1. Single file under one path — straightforward attribution
+ *   2. Multiple paths with overlapping prefixes — longest-prefix wins
+ *      (e.g. `events/active/E1/x.jpg` should attribute to
+ *       `events/active`, not `events`)
+ *   3. File outside any configured path — silently dropped, doesn't
+ *      throw or contaminate other buckets
+ *
+ * Tests exercise the EXPORTED side: write a backup_runs row via the
+ * service entry point and assert the statistics JSON shape. We don't
+ * stub `computePerPathStats` directly — the integration view is what
+ * the frontend actually consumes.
+ */
+
+const fs = require('fs');
+const path = require('path');
+
+const { bootCrmDb } = require('./helpers/crmDb');
+
+jest.setTimeout(120000);
+
+describe('backupService — per-Stage-B-path statistics', () => {
+  let db;
+  let cleanup;
+  let storagePath;
+  let backupService;
+
+  beforeAll(async () => {
+    ({ db, cleanup } = await bootCrmDb());
+    storagePath = process.env.STORAGE_PATH;
+    backupService = require('../../src/services/backupService');
+  }, 120000);
+
+  afterAll(async () => {
+    if (cleanup) await cleanup();
+  });
+
+  function mkFile(rel, content = 'x'.repeat(100)) {
+    const abs = path.join(storagePath, rel);
+    fs.mkdirSync(path.dirname(abs), { recursive: true });
+    fs.writeFileSync(abs, content);
+  }
+
+  beforeEach(async () => {
+    // Clean slate of any artefacts from prior tests
+    await db('backup_runs').del();
+    await db('app_settings').where('setting_type', 'backup').del();
+    await db('app_settings').insert([
+      { setting_key: 'backup_destination_type', setting_value: JSON.stringify('local'), setting_type: 'backup' },
+      { setting_key: 'backup_destination_path', setting_value: JSON.stringify(path.join(storagePath, 'destination')), setting_type: 'backup' },
+      { setting_key: 'backup_enabled', setting_value: JSON.stringify(true), setting_type: 'backup' },
+      { setting_key: 'backup_email_on_failure', setting_value: JSON.stringify(false), setting_type: 'backup' },
+      { setting_key: 'backup_include_archived', setting_value: JSON.stringify(true), setting_type: 'backup' },
+    ]).onConflict('setting_key').merge();
+    fs.mkdirSync(path.join(storagePath, 'destination'), { recursive: true });
+
+    // Restore canonical backup_paths from migration 109
+    const { DEFAULT_PATHS } = require('../../migrations/core/109_add_backup_paths');
+    await db('backup_paths').del();
+    await db('backup_paths').insert(DEFAULT_PATHS.map((row) => ({
+      ...row,
+      created_at: new Date(),
+      updated_at: new Date(),
+    })));
+
+    // Wipe leftover files between tests
+    for (const dir of ['events', 'business-docs', 'thumbnails', 'previews', 'heroes', 'uploads']) {
+      const p = path.join(storagePath, dir);
+      if (fs.existsSync(p)) fs.rmSync(p, { recursive: true, force: true });
+    }
+  });
+
+  it('attributes files to their owning backup_paths row', async () => {
+    mkFile('events/active/E1/photo-a.jpg', 'X'.repeat(1000));
+    mkFile('events/active/E1/photo-b.jpg', 'X'.repeat(2000));
+    mkFile('business-docs/quote/2026/Q-1.pdf', 'X'.repeat(500));
+    mkFile('thumbnails/E1/photo-a.jpg', 'X'.repeat(50));
+
+    // Disable the inline DB dump so we don't need pg_dump in tests;
+    // the file walker is what produces per_path.
+    await db('app_settings').insert({
+      setting_key: 'backup_database_inline_dump',
+      setting_value: JSON.stringify(false),
+      setting_type: 'backup',
+    }).onConflict('setting_key').merge();
+
+    // Seed a fake DB-backup row so the fail-loud guard is satisfied.
+    const fakeDump = path.join(storagePath, 'destination', 'fake.sql.gz');
+    fs.writeFileSync(fakeDump, 'pretend dump');
+    await db('database_backup_runs').insert({
+      started_at: new Date(),
+      completed_at: new Date(),
+      status: 'completed',
+      backup_type: 'pg',
+      file_path: fakeDump,
+      file_size_bytes: fs.statSync(fakeDump).size,
+      destination_path: fakeDump,
+    });
+
+    await backupService.runBackup(true);
+
+    const run = await db('backup_runs').orderBy('id', 'desc').first();
+    expect(run.status).toBe('completed');
+
+    const statsRaw = typeof run.statistics === 'string'
+      ? JSON.parse(run.statistics)
+      : run.statistics;
+    expect(statsRaw.per_path).toBeDefined();
+
+    // events/active should have 2 files (3000 bytes)
+    expect(statsRaw.per_path['events/active']).toEqual({ count: 2, size: 3000 });
+    // business-docs should have 1 file (500 bytes)
+    expect(statsRaw.per_path['business-docs']).toEqual({ count: 1, size: 500 });
+    // thumbnails should have 1 file (50 bytes)
+    expect(statsRaw.per_path['thumbnails']).toEqual({ count: 1, size: 50 });
+
+    // No spurious buckets for paths that had nothing
+    expect(statsRaw.per_path['previews']).toBeUndefined();
+    expect(statsRaw.per_path['heroes']).toBeUndefined();
+  });
+
+  it('archived path attributed separately from active when both have files', async () => {
+    mkFile('events/active/E1/active.jpg', 'X'.repeat(100));
+    mkFile('events/archived/E2/archived.jpg', 'X'.repeat(200));
+
+    // backup_include_archived already set true in beforeEach so the
+    // archived walker fires; same opt-out for inline DB dump.
+    await db('app_settings').insert({
+      setting_key: 'backup_database_inline_dump',
+      setting_value: JSON.stringify(false),
+      setting_type: 'backup',
+    }).onConflict('setting_key').merge();
+    const fakeDump = path.join(storagePath, 'destination', 'fake.sql.gz');
+    fs.writeFileSync(fakeDump, 'pretend dump');
+    await db('database_backup_runs').insert({
+      started_at: new Date(),
+      completed_at: new Date(),
+      status: 'completed',
+      backup_type: 'pg',
+      file_path: fakeDump,
+      file_size_bytes: fs.statSync(fakeDump).size,
+      destination_path: fakeDump,
+    });
+
+    await backupService.runBackup(true);
+
+    const run = await db('backup_runs').orderBy('id', 'desc').first();
+    const statsRaw = typeof run.statistics === 'string'
+      ? JSON.parse(run.statistics)
+      : run.statistics;
+
+    // events/active and events/archived attribute separately —
+    // longest-prefix match prevents `events/active/...` from claiming
+    // an `events/archived/...` file or vice versa.
+    expect(statsRaw.per_path['events/active']).toEqual({ count: 1, size: 100 });
+    expect(statsRaw.per_path['events/archived']).toEqual({ count: 1, size: 200 });
+  });
+});
+
+// NOTE on walker duplication
+//
+// If two `backup_paths` rows overlap (e.g. one row at `events` AND
+// another at `events/active`), the walker scans the same files twice
+// — once via each path. Per-path stats then attribute the file to the
+// longest-prefix-matching path BOTH times, producing inflated counts.
+//
+// The canonical seed in migration 109 contains no overlapping pairs,
+// so this isn't exercised in practice. But an admin who hand-adds a
+// broad row that overlaps an existing nested one will see double
+// counts in their next backup's statistics + the destination will
+// receive duplicate copies (wasting space). Worth flagging if anyone
+// reports it — the fix is to de-dupe `files` in
+// `getFilesToBackupInternal` before returning, OR to skip walking a
+// path if a longer one has already covered it.

backend/__tests__/integration/backupService.smoke.test.js

@@ -0,0 +1,143 @@
+/**
+ * Smoke tests for backupService's config resolution + file-collection
+ * and manifest validation paths — safety net ahead of the god-file
+ * decomposition.
+ *
+ * Uses the same real-SQLite harness as
+ * backupService.configurableWalker.test.js (bootCrmDb + a temp
+ * STORAGE_PATH) rather than the broken deep-mock approach in
+ * backupService.enhanced.test.js.
+ */
+
+const fs = require('fs');
+const path = require('path');
+
+const { bootCrmDb } = require('./helpers/crmDb');
+
+jest.setTimeout(120000);
+
+describe('backupService — config + file collection + manifest (smoke)', () => {
+  let db;
+  let cleanup;
+  let storagePath;
+  let backupService;
+  let backupManifest;
+
+  beforeAll(async () => {
+    ({ db, cleanup } = await bootCrmDb());
+    storagePath = process.env.STORAGE_PATH;
+    backupService = require('../../src/services/backupService');
+    backupManifest = require('../../src/services/backupManifest');
+  }, 120000);
+
+  afterAll(async () => {
+    if (cleanup) await cleanup();
+  });
+
+  beforeEach(async () => {
+    await db('app_settings').del();
+    // Reset the storage tree so each test starts from a pristine walk.
+    await fs.promises.rm(storagePath, { recursive: true, force: true });
+    await fs.promises.mkdir(storagePath, { recursive: true });
+  });
+
+  function seedFile(relPath, content = 'dummy bytes') {
+    const abs = path.join(storagePath, relPath);
+    fs.mkdirSync(path.dirname(abs), { recursive: true });
+    fs.writeFileSync(abs, content);
+    return abs;
+  }
+
+  async function insertBackupSetting(key, value) {
+    await db('app_settings').insert({
+      setting_key: key,
+      setting_value: value,
+      setting_type: 'backup',
+    });
+  }
+
+  describe('getBackupConfig', () => {
+    it('parses booleans, numbers, JSON arrays and plain strings from app_settings', async () => {
+      await insertBackupSetting('backup_enabled', 'true');
+      await insertBackupSetting('backup_include_archived', 'false');
+      await insertBackupSetting('backup_retention_days', '30');
+      await insertBackupSetting('backup_destination_path', '/backups/picpeak');
+      await insertBackupSetting('backup_email_recipients', '["a@example.com","b@example.com"]');
+      // Non-backup settings must not leak into the backup config.
+      await db('app_settings').insert({
+        setting_key: 'general_site_name',
+        setting_value: 'PicPeak',
+        setting_type: 'general',
+      });
+
+      const config = await backupService.getBackupConfig();
+
+      expect(config.backup_enabled).toBe(true);
+      expect(config.backup_include_archived).toBe(false);
+      expect(config.backup_retention_days).toBe(30);
+      expect(config.backup_destination_path).toBe('/backups/picpeak');
+      expect(config.backup_email_recipients).toEqual(['a@example.com', 'b@example.com']);
+      expect(config).not.toHaveProperty('general_site_name');
+      // Raw (unparsed) values are preserved on the non-enumerable __raw.
+      expect(String(config.__raw.backup_retention_days)).toBe('30');
+    });
+
+    it('returns an empty config object (not null) when nothing is configured', async () => {
+      const config = await backupService.getBackupConfig();
+      expect(config).not.toBeNull();
+      expect(Object.keys(config)).toHaveLength(0);
+    });
+  });
+
+  describe('getFilesToBackup', () => {
+    it('returns an empty list on a pristine storage tree', async () => {
+      const files = await backupService.getFilesToBackup({ backup_include_archived: true });
+      expect(files).toEqual([]);
+    });
+
+    it('captures path/relativePath/size/modified metadata for backed-up files', async () => {
+      const content = 'not really a jpeg';
+      const abs = seedFile('events/active/E9/pic.jpg', content);
+
+      const files = await backupService.getFilesToBackup({ backup_include_archived: true });
+      const entry = files.find((f) => f.relativePath === path.join('events/active/E9', 'pic.jpg'));
+
+      expect(entry).toBeDefined();
+      expect(entry.path).toBe(abs);
+      expect(entry.size).toBe(Buffer.byteLength(content));
+      // Not toBeInstanceOf(Date) — fs.stat mtime comes from a different
+      // realm under Jest and fails the cross-realm instanceof check.
+      expect(Object.prototype.toString.call(entry.modified)).toBe('[object Date]');
+    });
+  });
+
+  describe('validateBackupManifest', () => {
+    it('round-trips a generated manifest as valid', async () => {
+      seedFile('events/active/E1/a.jpg', 'aaa');
+      const files = await backupService.getFilesToBackup({ backup_include_archived: true });
+
+      const manifest = await backupManifest.generateManifest({
+        backupType: 'full',
+        backupPath: '/backup/run-1',
+        files,
+      });
+      const manifestPath = path.join(storagePath, 'manifest-smoke.json');
+      await backupManifest.saveManifest(manifest, manifestPath, 'json');
+
+      const result = await backupService.validateBackupManifest(manifestPath);
+      expect(result.valid).toBe(true);
+      expect(result.manifest.backup.type).toBe('full');
+      expect(result.manifest.files.count).toBe(files.length);
+      expect(result.manifest.verification.total_checksum).toBeTruthy();
+    });
+
+    it('flags a manifest missing required sections as invalid', async () => {
+      const badPath = path.join(storagePath, 'manifest-broken.json');
+      fs.writeFileSync(badPath, JSON.stringify({ manifest: { version: '2.0' } }));
+
+      const result = await backupService.validateBackupManifest(badPath);
+      expect(result.valid).toBe(false);
+      expect(result.error).toMatch(/Missing required section/);
+    });
+  });
+});

backend/__tests__/integration/bookingCutover.test.js

@@ -0,0 +1,170 @@
+/**
+ * Booking cutover — prepare_invoice's draft seam. convertToInvoiceOnly({draft})
+ * must create the invoice(s) but leave scheduled_send_at NULL so the scheduler
+ * never auto-sends them before the workflow's review gate + explicit
+ * send_document.
+ */
+const crypto = require('crypto');
+const { bootCrmDb, seedMinimal } = require('./helpers/crmDb');
+
+jest.setTimeout(120000);
+
+describe('booking cutover — draft invoices on hold', () => {
+  let db; let cleanup; let adminId; let customerId; let quoteService;
+
+  beforeAll(async () => {
+    ({ db, cleanup } = await bootCrmDb());
+    ({ adminId, customerId } = await seedMinimal(db));
+    quoteService = require('../../src/services/quoteService');
+  }, 120000);
+
+  afterAll(async () => { if (cleanup) await cleanup(); });
+
+  async function acceptedQuote() {
+    const dealUuid = crypto.randomUUID();
+    const [id] = await db('quotes').insert({
+      quote_number: `Q-${dealUuid.slice(0, 8)}`,
+      customer_account_id: customerId,
+      status: 'accepted',
+      currency: 'CHF',
+      issue_date: '2026-01-01',
+      net_amount_minor: 100000, vat_amount_minor: 0, shipping_amount_minor: 0, total_amount_minor: 100000,
+      // A non-delivery installment so the contrast (scheduled date vs null) is meaningful.
+      payment_term_snapshot: JSON.stringify({ installments: [{ percent: 100, trigger: 'quote_accepted', offset_days: 0, label: 'Total' }], net_days: 30 }),
+      deal_uuid: dealUuid,
+      created_by_admin_id: adminId,
+    });
+    return id;
+  }
+
+  it('draft mode creates the invoice with scheduled_send_at = NULL (held), and returns its id', async () => {
+    const quoteId = await acceptedQuote();
+    const res = await quoteService.convertToInvoiceOnly(quoteId, adminId, { draft: true });
+    expect(Array.isArray(res.invoiceIds)).toBe(true);
+    expect(res.invoiceIds.length).toBeGreaterThanOrEqual(1);
+
+    const inv = await db('invoices').where({ id: res.invoiceIds[0] }).first();
+    expect(inv.status).toBe('scheduled');     // editable + sendInvoice can issue it
+    expect(inv.scheduled_send_at == null).toBe(true); // held — scheduler won't auto-send
+  });
+
+  it('without draft, the same installment IS scheduled (scheduled_send_at set)', async () => {
+    const quoteId = await acceptedQuote();
+    const res = await quoteService.convertToInvoiceOnly(quoteId, adminId);
+    const inv = await db('invoices').where({ id: res.invoiceIds[0] }).first();
+    expect(inv.status).toBe('scheduled');
+    expect(inv.scheduled_send_at == null).toBe(false); // normal convert → auto-send date set
+  });
+
+  it('prepare_event path (convertToEvent hold) creates a DRAFT event with held invoices', async () => {
+    const quoteId = await acceptedQuote();
+    const res = await quoteService.convertToEvent(quoteId, adminId, { hold: true });
+    expect(res.eventId).toBeGreaterThanOrEqual(1);
+    expect(Array.isArray(res.invoiceIds)).toBe(true);
+    expect(res.invoiceIds.length).toBeGreaterThanOrEqual(1);
+
+    const ev = await db('events').where({ id: res.eventId }).first();
+    expect(ev.is_draft == true || ev.is_draft === 1).toBe(true); // created as a draft gallery
+
+    // Every invoice the event scheduled is held (no auto-send before the gate).
+    const invs = await db('invoices').whereIn('id', res.invoiceIds);
+    for (const inv of invs) expect(inv.scheduled_send_at == null).toBe(true);
+
+    // Quote is now linked to the event — convertToInvoiceOnly must NOT be called
+    // again for it (the flow's prepare_invoice adopts these ids instead).
+    const q = await db('quotes').where({ id: quoteId }).first();
+    expect(q.converted_event_id).toBe(res.eventId);
+  });
+
+  it('draft mode with the DEFAULT (after_delivery) payment term yields a SENDABLE scheduled invoice, not pending_delivery', async () => {
+    // Reproduces the booking_invoice_only flow on a quote with no explicit
+    // payment timing: the default installment is after_delivery, which would
+    // otherwise be pending_delivery — a status sendInvoice (send_document) rejects.
+    const dealUuid = crypto.randomUUID();
+    const [quoteId] = await db('quotes').insert({
+      quote_number: `Q-${dealUuid.slice(0, 8)}`,
+      customer_account_id: customerId,
+      status: 'accepted',
+      currency: 'CHF',
+      issue_date: '2026-01-01',
+      net_amount_minor: 50000, vat_amount_minor: 0, shipping_amount_minor: 0, total_amount_minor: 50000,
+      // No payment_term_snapshot → spawnInstallmentInvoices falls back to a single
+      // 100% after_delivery installment.
+      deal_uuid: dealUuid,
+      created_by_admin_id: adminId,
+    });
+    const res = await quoteService.convertToInvoiceOnly(quoteId, adminId, { draft: true });
+    const inv = await db('invoices').where({ id: res.invoiceIds[0] }).first();
+    expect(inv.status).toBe('scheduled');            // sendInvoice accepts this
+    expect(inv.scheduled_send_at == null).toBe(true); // still held — no auto-send
+  });
+
+  it('finalizeQuoteResponses only fires once the 15-min response window has locked', async () => {
+    const mk = async (lockOffsetMs) => {
+      const dealUuid = crypto.randomUUID();
+      const [id] = await db('quotes').insert({
+        quote_number: `Q-${dealUuid.slice(0, 8)}`,
+        customer_account_id: customerId,
+        status: 'accepted',
+        currency: 'CHF', issue_date: '2026-01-01',
+        net_amount_minor: 1000, vat_amount_minor: 0, shipping_amount_minor: 0, total_amount_minor: 1000,
+        responded_at: new Date().toISOString(),
+        response_locked_at: new Date(Date.now() + lockOffsetMs).toISOString(),
+        accepted_at: new Date().toISOString(),
+        deal_uuid: dealUuid,
+        created_by_admin_id: adminId,
+      });
+      return id;
+    };
+    const openId = await mk(15 * 60 * 1000);   // still inside the window
+    const lockedId = await mk(-60 * 1000);      // window already closed
+
+    const emitted = await quoteService.finalizeQuoteResponses();
+    expect(emitted).toBeGreaterThanOrEqual(1);
+
+    const open = await db('quotes').where({ id: openId }).first();
+    const locked = await db('quotes').where({ id: lockedId }).first();
+    expect(open.workflow_response_emitted_at == null).toBe(true);   // deferred — not yet fired
+    expect(locked.workflow_response_emitted_at == null).toBe(false); // fired + stamped
+
+    // Idempotent: a second sweep doesn't re-fire the already-stamped one.
+    const again = await db('quotes').where({ id: lockedId })
+      .whereNull('workflow_response_emitted_at').update({ workflow_response_emitted_at: new Date() });
+    expect(again).toBe(0);
+  });
+
+  it('reserve_date path (convertToEvent skipInvoices) creates a draft event with NO invoices', async () => {
+    const quoteId = await acceptedQuote();
+    const res = await quoteService.convertToEvent(quoteId, adminId, { hold: true, skipInvoices: true });
+    expect(res.eventId).toBeGreaterThanOrEqual(1);
+    expect(res.invoiceIds).toEqual([]);
+    const invCount = await db('invoices').where({ event_id: res.eventId }).count({ c: '*' }).first();
+    expect(Number(invCount.c)).toBe(0); // pure date hold — no money documents
+  });
+
+  it('prepare_quote path (duplicateQuote) creates a new DRAFT quote — no in-trx deadlock', async () => {
+    const quoteId = await acceptedQuote();
+    const newId = await quoteService.duplicateQuote(quoteId, adminId);
+    expect(newId).toBeGreaterThanOrEqual(1);
+    expect(newId).not.toBe(quoteId);
+    const q = await db('quotes').where({ id: newId }).first();
+    expect(q.status).toBe('draft');
+  });
+
+  it('registers prepare_gallery / reserve_date / prepare_quote as real actions', () => {
+    const { registry } = require('../../src/services/workflows'); // loads actions.js (side-effect registration)
+    for (const a of ['prepare_gallery', 'reserve_date', 'prepare_quote', 'prepare_event', 'prepare_invoice', 'send_document']) {
+      expect(typeof registry.getAction(a)).toBe('function');
+    }
+  });
+
+  it('prepare_contract path (createFromQuote) completes under SQLite — no in-trx deadlock', async () => {
+    const contractService = require('../../src/services/contractService');
+    const quoteId = await acceptedQuote();
+    const res = await contractService.createFromQuote(quoteId, adminId);
+    expect(res.contractId).toBeGreaterThanOrEqual(1);
+    expect(res.alreadyConverted).toBe(false);
+    const c = await db('contracts').where({ id: res.contractId }).first();
+    expect(c).toBeTruthy();
+  });
+});

backend/__tests__/integration/businessProfileLogoUpload.test.js

@@ -0,0 +1,195 @@
+/**
+ * POST /api/admin/business-profile/logo and PUT /api/admin/business-profile
+ * — GHSA-6wrv-9pr4-hhmw regression coverage.
+ *
+ * The upload route used to take the stored file extension straight from
+ * the client-supplied filename and only checked `file.mimetype` against an
+ * allowlist — a file could declare an image MIME type while carrying a
+ * `.html`/`.js` extension and arbitrary content, land in the same-origin
+ * `/uploads/logos` static mount, and execute as script. The mass-assignable
+ * `logoPath` field on PUT compounded it: an attacker could point the
+ * "logo" at any other uploaded file.
+ *
+ * These tests pin:
+ *   (a) a MIME/extension mismatch is rejected at upload,
+ *   (b) the extension actually written to disk always matches the
+ *       validated MIME type, never the client-supplied filename,
+ *   (c) legitimate PNG/JPEG/SVG uploads still succeed,
+ *   (d) `logoPath` on PUT cannot be set to an arbitrary string pointing at
+ *       another file, only to a path the upload route itself produced.
+ *
+ * Defense-in-depth (not a re-opening of the above): fileFilter only pairs
+ * the claimed MIME type against the extension — it can't see the bytes,
+ * since it runs before multer finishes writing the stream to disk. A file
+ * whose declared MIME/extension pair is valid but whose actual content
+ * doesn't match (e.g. a PNG-declared upload that isn't really a PNG) is
+ * now caught by validateFileContent() (magic-number check) after multer
+ * writes it, closing the gap where declared-vs-actual content diverges.
+ */
+
+const path = require('path');
+const fs = require('fs');
+const os = require('os');
+
+const tmpDir = fs.mkdtempSync(path.join(os.tmpdir(), 'picpeak-bplogo-test-'));
+process.env.NODE_ENV = 'test';
+process.env.TEST_DATABASE_PATH = path.join(tmpDir, 'db.sqlite');
+process.env.STORAGE_PATH = path.join(tmpDir, 'storage');
+fs.mkdirSync(process.env.STORAGE_PATH, { recursive: true });
+process.env.JWT_SECRET = process.env.JWT_SECRET || 'bplogo-route-test-secret';
+
+const request = require('supertest');
+const {
+  bootCrmDb, seedMinimal, assignAdminRole, mintAdminToken, buildRouteApp,
+} = require('./helpers/crmDb');
+
+// Real magic-number-prefixed payloads, for content-sniffing to accept.
+const REAL_PNG_BYTES = Buffer.concat([
+  Buffer.from([0x89, 0x50, 0x4E, 0x47, 0x0D, 0x0A, 0x1A, 0x0A]),
+  Buffer.from('not a real png body, but the header is real'),
+]);
+const REAL_JPEG_BYTES = Buffer.concat([
+  Buffer.from([0xFF, 0xD8, 0xFF]),
+  Buffer.from('not a real jpeg body, but the header is real'),
+]);
+
+describe('business profile — logo upload content/extension validation', () => {
+  let db;
+  let cleanup;
+  let app;
+  let token;
+
+  const uploadLogo = (buffer, filename, mimetype) => request(app)
+    .post('/api/admin/business-profile/logo')
+    .set('Authorization', `Bearer ${token}`)
+    .attach('logo', buffer, { filename, contentType: mimetype });
+
+  const put = (payload) => request(app)
+    .put('/api/admin/business-profile')
+    .set('Authorization', `Bearer ${token}`)
+    .send(payload);
+
+  const get = () => request(app)
+    .get('/api/admin/business-profile')
+    .set('Authorization', `Bearer ${token}`);
+
+  const profileOf = (res) => (res.body.data || res.body).profile;
+
+  beforeAll(async () => {
+    ({ db, cleanup } = await bootCrmDb());
+    const { adminId } = await seedMinimal(db);
+    await assignAdminRole(db, adminId, 'super_admin');
+    token = mintAdminToken(adminId);
+    app = buildRouteApp('/api/admin/business-profile', require('../../src/routes/adminBusinessProfile'));
+  }, 120000);
+
+  afterAll(async () => {
+    if (cleanup) await cleanup();
+  });
+
+  // fileFilter rejections surface via Express's generic error handler
+  // (the pre-existing behaviour of every sibling logo/favicon upload
+  // route in this codebase — none of them special-case multer's
+  // fileFilter `Error` into a 400 either), so the status code itself
+  // can be 400 or 500 depending on environment. What actually matters
+  // for GHSA-6wrv-9pr4-hhmw is that the request never succeeds and
+  // nothing with the dangerous extension is ever written to disk.
+  const logosDirFiles = () => {
+    const logosDir = path.join(process.env.STORAGE_PATH, 'uploads', 'logos');
+    return fs.existsSync(logosDir) ? fs.readdirSync(logosDir) : [];
+  };
+
+  it('rejects an HTML/script payload disguised as an image via mismatched extension', async () => {
+    const evil = Buffer.from('<script>alert(document.domain)</script>');
+    const res = await uploadLogo(evil, 'evil.html', 'image/svg+xml');
+    expect(res.status).not.toBe(200);
+    expect(logosDirFiles().some((f) => f.endsWith('.html'))).toBe(false);
+  });
+
+  it('rejects a .js file disguised with an image MIME type', async () => {
+    const evil = Buffer.from('alert(1)');
+    const res = await uploadLogo(evil, 'evil.js', 'image/png');
+    expect(res.status).not.toBe(200);
+    expect(logosDirFiles().some((f) => f.endsWith('.js'))).toBe(false);
+  });
+
+  it('rejects a disallowed MIME type outright', async () => {
+    const res = await uploadLogo(Buffer.from('whatever'), 'file.pdf', 'application/pdf');
+    expect(res.status).not.toBe(200);
+    expect(logosDirFiles().some((f) => f.endsWith('.pdf'))).toBe(false);
+  });
+
+  it('accepts a legitimate PNG upload and stores it with a .png extension', async () => {
+    const res = await uploadLogo(REAL_PNG_BYTES, 'logo.png', 'image/png');
+    expect(res.status).toBe(200);
+    const logoPath = (res.body.data || res.body).logoPath;
+    expect(logoPath).toMatch(/^\/uploads\/logos\/pdf-logo-\d+\.png$/);
+
+    const onDisk = path.join(process.env.STORAGE_PATH, logoPath.replace(/^\//, ''));
+    expect(fs.existsSync(onDisk)).toBe(true);
+
+    expect(profileOf(await get()).logoPath).toBe(logoPath);
+  });
+
+  it('accepts a legitimate JPEG upload and stores it with a .jpg extension', async () => {
+    const res = await uploadLogo(REAL_JPEG_BYTES, 'logo.jpg', 'image/jpeg');
+    expect(res.status).toBe(200);
+    const logoPath = (res.body.data || res.body).logoPath;
+    expect(logoPath).toMatch(/^\/uploads\/logos\/pdf-logo-\d+\.jpg$/);
+  });
+
+  it('rejects a PNG-declared upload whose bytes are not actually a PNG, and leaves nothing on disk', async () => {
+    const before = logosDirFiles();
+    const res = await uploadLogo(Buffer.from('totally not a png'), 'logo.png', 'image/png');
+    expect(res.status).toBe(400);
+    expect(res.body.error).toMatch(/content does not match/i);
+
+    // No new file left behind: the rejected upload's own file was cleaned
+    // up, and every other file on disk (if any) is unchanged.
+    expect(logosDirFiles()).toEqual(before);
+  });
+
+  it('accepts a legitimate SVG upload and always stores it with a .svg extension, even under a spoofed filename', async () => {
+    const svg = Buffer.from('<svg xmlns="http://www.w3.org/2000/svg"><rect width="1" height="1"/></svg>');
+    // Client-declared filename ext is .svg here to pass validateFileType
+    // (mismatched ext is covered by the rejection tests above); the point
+    // of this test is that the ON-DISK extension comes from the MIME type
+    // lookup table, not path.extname(originalname).
+    const res = await uploadLogo(svg, 'vector-logo.svg', 'image/svg+xml');
+    expect(res.status).toBe(200);
+    const logoPath = (res.body.data || res.body).logoPath;
+    expect(logoPath).toMatch(/^\/uploads\/logos\/pdf-logo-\d+\.svg$/);
+  });
+
+  it('rejects logoPath on PUT set to an arbitrary string pointing at another file', async () => {
+    const before = profileOf(await get()).logoPath;
+
+    const res = await put({ logoPath: '/uploads/logos/cms-somepage-1234.png' });
+    expect(res.status).toBe(400);
+
+    expect(profileOf(await get()).logoPath).toBe(before);
+  });
+
+  it('rejects logoPath on PUT with a path-traversal payload', async () => {
+    const res = await put({ logoPath: '/uploads/logos/../../../../etc/passwd' });
+    expect(res.status).toBe(400);
+  });
+
+  it('accepts logoPath on PUT when it matches the pattern this route itself writes', async () => {
+    const upload = await uploadLogo(REAL_PNG_BYTES, 'logo2.png', 'image/png');
+    const uploadedPath = (upload.body.data || upload.body).logoPath;
+
+    // Round-trip: PUT-ing back the exact value the upload endpoint
+    // returned (what the frontend's generic profile save does) must
+    // keep working.
+    const res = await put({ logoPath: uploadedPath });
+    expect(res.status).toBe(200);
+    expect(profileOf(await get()).logoPath).toBe(uploadedPath);
+  });
+
+  it('still allows clearing logoPath with an empty string', async () => {
+    const res = await put({ logoPath: '' });
+    expect(res.status).toBe(200);
+    expect(profileOf(await get()).logoPath).toBe('');
+  });
+});

backend/__tests__/integration/captureDateBackfill.test.js

@@ -0,0 +1,227 @@
+/**
+ * Backfilling captured_at on a library imported before #1172.
+ *
+ * The point of the endpoint, rather than a migration: it resolves originals
+ * through resolvePhotoFilePath, which is the only path that reaches an
+ * external row. The thumbnail regenerator resolves under
+ * storage/events/active/<photo.path>, which never exists for those (#1129) —
+ * so it cannot be the model.
+ */
+
+const fs = require('fs');
+const path = require('path');
+const os = require('os');
+const express = require('express');
+const request = require('supertest');
+const sharp = require('sharp');
+
+describe('capture date backfill (#1172)', () => {
+  let tmpDir; let db; let app; let mediaRoot;
+
+  const writeJpegWithExif = async (abs, iso) => {
+    await fs.promises.mkdir(path.dirname(abs), { recursive: true });
+    const d = new Date(iso);
+    const pad = (n) => String(n).padStart(2, '0');
+    const exifDate = `${d.getUTCFullYear()}:${pad(d.getUTCMonth() + 1)}:${pad(d.getUTCDate())} `
+      + `${pad(d.getUTCHours())}:${pad(d.getUTCMinutes())}:${pad(d.getUTCSeconds())}`;
+    await sharp({ create: { width: 60, height: 40, channels: 3, background: { r: 9, g: 9, b: 9 } } })
+      .withExif({ IFD2: { DateTimeOriginal: exifDate } }).jpeg().toFile(abs);
+  };
+
+  const settle = async () => { for (let i = 0; i < 60; i++) { await new Promise((r) => setTimeout(r, 50)); const s = await status(); if (!s.body.isRunning) return s; } throw new Error('backfill did not settle'); };
+  const status = () => request(app).get('/api/admin/photos/repair-capture-dates/status');
+
+  beforeAll(async () => {
+    tmpDir = await fs.promises.mkdtemp(path.join(os.tmpdir(), 'picpeak-capfill-'));
+    mediaRoot = path.join(tmpDir, 'media');
+    await fs.promises.mkdir(mediaRoot, { recursive: true });
+
+    process.env.NODE_ENV = 'test';
+    process.env.TEST_DATABASE_PATH = path.join(tmpDir, 'data', 'db.sqlite');
+    await fs.promises.mkdir(path.dirname(process.env.TEST_DATABASE_PATH), { recursive: true });
+    process.env.STORAGE_PATH = path.join(tmpDir, 'storage');
+    process.env.EXTERNAL_MEDIA_ROOT = mediaRoot;
+    process.env.JWT_SECRET = process.env.JWT_SECRET || 'capfill-secret';
+
+    jest.resetModules();
+    jest.doMock('../../src/middleware/auth', () => ({
+      adminAuth: (req, _res, next) => { req.admin = { id: 1, username: 'tester', roleName: 'admin' }; next(); },
+    }));
+    jest.doMock('../../src/middleware/permissions', () => ({
+      requirePermission: () => (_req, _res, next) => next(),
+    }));
+    jest.doMock('../../src/utils/logger', () => ({
+      debug: jest.fn(), info: jest.fn(), warn: jest.fn(), error: jest.fn(),
+    }));
+
+    ({ db } = await require('./helpers/crmDb').bootCrmDb());
+
+    app = express();
+    app.use(express.json());
+    app.use('/api/admin/photos', require('../../src/routes/adminPhotoDimensions'));
+  }, 180000);
+
+  afterAll(async () => {
+    if (db) await db.destroy?.();
+    await fs.promises.rm(tmpDir, { recursive: true, force: true }).catch(() => {});
+  });
+
+  async function seed({ relpath, exifIso, writeFile = true, archived = false }) {
+    await db('photos').del();
+    await db('events').del();
+    const [e] = await db('events').insert({
+      slug: 'capfill', event_type: 'wedding', event_name: 'capfill', event_date: '2026-01-01',
+      host_email: 'h@example.com', admin_email: 'a@example.com', password_hash: 'x',
+      share_link: `capfill-${Math.random()}`, expires_at: new Date().toISOString(),
+      source_mode: 'reference', external_path: 'trip', is_archived: archived,
+    }).returning('id');
+    const eventId = typeof e === 'object' ? e.id : e;
+    if (writeFile) await writeJpegWithExif(path.join(mediaRoot, 'trip', relpath), exifIso);
+    const [p] = await db('photos').insert({
+      event_id: eventId, filename: path.basename(relpath), path: `capfill/${path.basename(relpath)}`,
+      // Root-relative, as this branch stores it (#1163) — the file lives at
+      // <mediaRoot>/trip/<relpath>.
+      type: 'individual', source_origin: 'external', external_relpath: `trip/${relpath}`,
+      uploaded_at: new Date().toISOString(), captured_at: null,
+    }).returning('id');
+    return { eventId, photoId: typeof p === 'object' ? p.id : p };
+  }
+
+  it('fills captured_at for an external photo the thumbnail regenerator cannot reach', async () => {
+    const { photoId } = await seed({ relpath: 'a.jpg', exifIso: '2026-06-01T09:45:03Z' });
+
+    const res = await request(app).post('/api/admin/photos/repair-capture-dates');
+    expect(res.status).toBe(200);
+    expect(res.body.count).toBe(1);
+    const done = await settle();
+
+    expect(done.body.lastResult.success).toBe(1);
+    expect((await db('photos').where({ id: photoId }).first()).captured_at).toBeTruthy();
+  });
+
+  it('counts a photo with no EXIF separately from a failure', async () => {
+    // "The mount is broken" and "these files carry no date" need different
+    // answers from an operator, so they are not the same number.
+    await db('photos').del(); await db('events').del();
+    const { photoId } = await seed({ relpath: 'plain.jpg', exifIso: '2026-06-01T09:45:03Z', writeFile: false });
+    await sharp({ create: { width: 40, height: 30, channels: 3, background: { r: 1, g: 1, b: 1 } } })
+      .jpeg().toFile(path.join(mediaRoot, 'trip', 'plain.jpg'));
+
+    await request(app).post('/api/admin/photos/repair-capture-dates');
+    const done = await settle();
+
+    expect(done.body.lastResult).toMatchObject({ success: 0, noExif: 1, failed: 0 });
+    expect((await db('photos').where({ id: photoId }).first()).captured_at).toBeNull();
+  });
+
+  it('counts an unreachable original as a failure, not as missing EXIF', async () => {
+    await seed({ relpath: 'gone.jpg', exifIso: '2026-06-01T09:45:03Z', writeFile: false });
+
+    await request(app).post('/api/admin/photos/repair-capture-dates');
+    const done = await settle();
+
+    expect(done.body.lastResult).toMatchObject({ success: 0, noExif: 0, failed: 1 });
+  });
+
+  it('reports nothing to do once every photo has a date', async () => {
+    const { photoId } = await seed({ relpath: 'b.jpg', exifIso: '2026-06-02T09:00:00Z' });
+    await db('photos').where({ id: photoId }).update({ captured_at: new Date().toISOString() });
+
+    const res = await request(app).post('/api/admin/photos/repair-capture-dates');
+
+    expect(res.body.count).toBe(0);
+    expect((await status()).body.withoutCaptureDate).toBe(0);
+  });
+
+  it('skips a watcher-imported video, which carries media_type "image"', async () => {
+    // fileWatcher.processNewPhoto sets type='video' and a video/* mime but
+    // never media_type (fileWatcher.js:128-130), so the row keeps the 'image'
+    // default from migration 048. Filtering on media_type alone queued it every
+    // run: extractCaptureDate returns null for a video, captured_at stays null,
+    // and the backlog never cleared.
+    const { eventId } = await seed({ relpath: 'clip.jpg', exifIso: '2026-06-01T09:45:03Z', writeFile: false });
+    await db('photos').del();
+    await db('photos').insert({
+      event_id: eventId, filename: 'clip.mp4', path: 'capfill/clip.mp4',
+      type: 'video', media_type: 'image', mime_type: 'video/mp4',
+      source_origin: 'external', external_relpath: 'trip/clip.mp4',
+      uploaded_at: new Date().toISOString(), captured_at: null,
+    });
+
+    const res = await request(app).post('/api/admin/photos/repair-capture-dates');
+    expect(res.body.count).toBe(0);
+
+    const s = await status();
+    // And it is not counted as a permanent backlog either.
+    expect(s.body.total).toBe(0);
+    expect(s.body.withoutCaptureDate).toBe(0);
+  });
+
+  it('never reports more dated photos than it has photos', async () => {
+    // Both counts come from one aggregate; as two queries an import committing
+    // between them produced withCaptureDate > total and a negative backlog.
+    const { photoId } = await seed({ relpath: 'counted.jpg', exifIso: '2026-06-05T08:00:00Z' });
+    await db('photos').where({ id: photoId }).update({ captured_at: new Date().toISOString() });
+
+    const s = await status();
+    expect(s.body.total).toBe(1);
+    expect(s.body.withCaptureDate).toBe(1);
+    expect(s.body.withoutCaptureDate).toBe(0);
+    expect(s.body.withoutCaptureDate).toBeGreaterThanOrEqual(0);
+  });
+
+  it('skips archived events instead of failing them on every run', async () => {
+    // Archiving deletes the originals and keeps the rows, so an archived photo
+    // can never get a date. Counting it would fail it every pass and leave the
+    // status endpoint permanently reporting a backlog.
+    await seed({ relpath: 'archived.jpg', exifIso: '2026-06-04T09:00:00Z', archived: true });
+
+    const res = await request(app).post('/api/admin/photos/repair-capture-dates');
+
+    expect(res.body.count).toBe(0);
+    const s = await status();
+    expect(s.body.total).toBe(0);
+    expect(s.body.withoutCaptureDate).toBe(0);
+    expect(s.body.isRunning).toBe(false);
+  });
+
+  it('does not overwrite a date written while it was running', async () => {
+    // whereNull on the update: an import or a replacement finishing mid-run has
+    // already written a better value than this pass would.
+    const { photoId } = await seed({ relpath: 'c.jpg', exifIso: '2026-06-03T09:00:00Z' });
+    const claimed = '2020-01-01T00:00:00.000Z';
+
+    const res = await request(app).post('/api/admin/photos/repair-capture-dates');
+    expect(res.body.count).toBe(1);
+    await db('photos').where({ id: photoId }).update({ captured_at: claimed });
+    const done = await settle();
+
+    expect(new Date((await db('photos').where({ id: photoId }).first()).captured_at).toISOString()).toBe(claimed);
+    expect(done.body.lastResult.success).toBe(0);
+    // Read but not written, so it is accounted for rather than dropped.
+    expect(done.body.lastResult.skipped).toBe(1);
+  });
+
+  it('does not date a row whose file was replaced while it was reading (#1201)', async () => {
+    // replacePhoto swaps a NEW file under an existing row and rewrites
+    // path/filename (reachable from replace_by_name). The replacement carries
+    // no date of its own, so captured_at is still NULL and the whereNull guard
+    // alone would let the previous file's EXIF date land on it. The write is
+    // fenced on the identity that was read, so the row is skipped instead —
+    // and not counted as updated either.
+    const { photoId } = await seed({ relpath: 'orig.jpg', exifIso: '2026-06-03T09:00:00Z' });
+
+    const res = await request(app).post('/api/admin/photos/repair-capture-dates');
+    expect(res.body.count).toBe(1);
+    // Simulate the replacement landing before the loop writes.
+    await db('photos').where({ id: photoId })
+      .update({ path: 'capfill/replaced.jpg', filename: 'replaced.jpg' });
+    const done = await settle();
+
+    expect((await db('photos').where({ id: photoId }).first()).captured_at).toBeNull();
+    expect(done.body.lastResult.success).toBe(0);
+    // Not an error and not "no EXIF" — the date was found, another writer just
+    // got there first. It stays in the backlog for the next run.
+    expect(done.body.lastResult).toMatchObject({ noExif: 0, failed: 0, skipped: 1 });
+  });
+});

backend/__tests__/integration/crmSchema.test.js

@@ -0,0 +1,116 @@
+/**
+ * Schema-shape regression net for the CRM consolidated migration.
+ *
+ * Pins the table/column layout that the route + service layer expect
+ * after `migrations/core/107_crm_consolidated.js` runs. The schema-
+ * drift workflow (#530) catches Postgres-only FK ordering bugs (the
+ * forward-reference deferral added in this PR), but it doesn't notice
+ * if a future edit silently drops a column the service code reads —
+ * SQLite would just return undefined and the broken behavior would
+ * land on beta.
+ *
+ * Touches the lineage chain (deal_uuid + back-pointer FKs) explicitly
+ * so a rename or removal there fails the test instead of silently
+ * breaking the lineage card.
+ */
+
+const { bootCrmDb, seedMinimal } = require('./helpers/crmDb');
+
+describe('CRM schema after core migrations', () => {
+  let db;
+  let cleanup;
+
+  beforeAll(async () => {
+    ({ db, cleanup } = await bootCrmDb());
+  }, 120000);
+
+  afterAll(async () => {
+    if (cleanup) await cleanup();
+  });
+
+  describe('table layout', () => {
+    const expectedTables = [
+      'admin_users', 'customer_accounts', 'business_profile', 'business_bank_accounts',
+      'events', 'document_sequences',
+      'quotes', 'quote_line_items', 'quote_line_item_presets', 'quote_action_tokens',
+      'contracts', 'contract_blocks', 'contract_block_inclusions', 'contract_action_tokens',
+      'invoices', 'invoice_line_items', 'invoice_payment_log', 'invoice_payment_check_tokens',
+      'customer_hour_entries',
+      'payment_term_templates', 'payment_net_days_templates', 'payment_timing_templates',
+      'event_payment_plans',
+    ];
+
+    it.each(expectedTables)('has table %s', async (table) => {
+      expect(await db.schema.hasTable(table)).toBe(true);
+    });
+  });
+
+  describe('deal_uuid lineage columns', () => {
+    // Every document in one engagement shares a deal_uuid — the
+    // lineage card joins on it. Drop the column anywhere in the chain
+    // and the card silently returns partial data.
+    it.each(['quotes', 'contracts', 'invoices'])(
+      '%s has deal_uuid column',
+      async (table) => {
+        expect(await db.schema.hasColumn(table, 'deal_uuid')).toBe(true);
+      }
+    );
+
+    // The back-pointer FKs were the source of the schema-drift bug
+    // we fixed in this PR (forward references). Pin them.
+    it('quotes has converted_contract_id back-pointer', async () => {
+      expect(await db.schema.hasColumn('quotes', 'converted_contract_id')).toBe(true);
+    });
+    it('invoices has source_contract_id back-pointer', async () => {
+      expect(await db.schema.hasColumn('invoices', 'source_contract_id')).toBe(true);
+    });
+    it('invoices has source_quote_id back-pointer', async () => {
+      expect(await db.schema.hasColumn('invoices', 'source_quote_id')).toBe(true);
+    });
+  });
+
+  describe('Storno discriminator columns', () => {
+    // kind='storno' + cancels_invoice_id + negative totals are the
+    // shape every aggregate filter relies on (feedback_storno_filter_
+    // everywhere). Pin the columns so a rename doesn't silently break
+    // every revenue report.
+    it('invoices has kind discriminator', async () => {
+      expect(await db.schema.hasColumn('invoices', 'kind')).toBe(true);
+    });
+    it('invoices has cancels_invoice_id self-ref', async () => {
+      expect(await db.schema.hasColumn('invoices', 'cancels_invoice_id')).toBe(true);
+    });
+    it('invoices has replaces_invoice_id self-ref', async () => {
+      expect(await db.schema.hasColumn('invoices', 'replaces_invoice_id')).toBe(true);
+    });
+  });
+
+  describe('Event time columns (migration 137)', () => {
+    // The admin calendar reads these to render timed vs. full-day
+    // tiles. Per the feedback_migration_preserve_visuals rule, the
+    // default has to be `is_full_day=true` so existing rows keep
+    // their pre-migration visual.
+    it('events has event_time_start', async () => {
+      expect(await db.schema.hasColumn('events', 'event_time_start')).toBe(true);
+    });
+    it('events has event_time_end', async () => {
+      expect(await db.schema.hasColumn('events', 'event_time_end')).toBe(true);
+    });
+    it('events has is_full_day', async () => {
+      expect(await db.schema.hasColumn('events', 'is_full_day')).toBe(true);
+    });
+  });
+
+  describe('seed paths', () => {
+    it('admin + customer seed inserts cleanly', async () => {
+      const { adminId, customerId } = await seedMinimal(db);
+      expect(adminId).toBeTruthy();
+      expect(customerId).toBeTruthy();
+
+      const admin = await db('admin_users').where({ id: adminId }).first();
+      const customer = await db('customer_accounts').where({ id: customerId }).first();
+      expect(admin.email).toBe('tester@example.com');
+      expect(customer.email).toBe('customer@example.com');
+    });
+  });
+});

backend/__tests__/integration/discountLineItems.test.js

@@ -0,0 +1,79 @@
+/**
+ * Negative line items (Rabatt / manual discount lines) are accepted
+ * end-to-end as long as the resulting total stays ≥ 0. When the
+ * discount would drive the total negative, the service rejects with
+ * a clear, code-tagged error so the admin is steered to Storno for
+ * credit-note workflows.
+ *
+ * Touches the actual createInvoice / createQuote service paths so a
+ * future change to either computeTotals or the guard fires this test.
+ */
+
+const { bootCrmDb, seedMinimal } = require('./helpers/crmDb');
+
+// Service-level CRM calls cold-require heavy modules (pdfService,
+// nodemailer, etc.) on first use; the global 5 s per-test budget is
+// too tight for that. Bump it for this file only.
+jest.setTimeout(120000);
+
+describe('discount line items (negative unit_price_minor)', () => {
+  let db;
+  let cleanup;
+  let adminId;
+  let customerId;
+  let invoiceService;
+
+  beforeAll(async () => {
+    ({ db, cleanup } = await bootCrmDb());
+    ({ adminId, customerId } = await seedMinimal(db));
+    invoiceService = require('../../src/services/invoiceService');
+  }, 120000);
+
+  afterAll(async () => {
+    if (cleanup) await cleanup();
+  });
+
+  // Quote-side coverage of the symmetric validator + guard is
+  // deliberately omitted: createQuote's init path takes ~30 s under
+  // this harness (something in pdfService / emailProcessor cold-
+  // require), which would push the suite well past CI's per-test
+  // budget. The shape of the guard is identical to the invoice one
+  // covered below; a future change to extract the slow init or to
+  // stub it for tests should re-enable a parallel quote test.
+
+  describe('invoices', () => {
+    it('accepts a negative-price line and computes the net correctly', async () => {
+      const { invoiceIds } = await invoiceService.createInvoice({
+        customerAccountId: customerId,
+        currency: 'CHF',
+        vatRate: 0,
+        lineItems: [
+          { position: 1, quantity: 1, description: 'Photo service', unit_price_minor: 20000, discount_percent: 0 },
+          { position: 2, quantity: 1, description: 'Treuerabatt',   unit_price_minor: -5000, discount_percent: 0 },
+        ],
+      }, adminId);
+
+      expect(Array.isArray(invoiceIds)).toBe(true);
+      expect(invoiceIds.length).toBe(1);
+
+      const row = await db('invoices').where({ id: invoiceIds[0] }).first();
+      expect(row.net_amount_minor).toBe(15000);
+      expect(row.total_amount_minor).toBe(15000);
+    });
+
+    it('rejects when the discount drives the total negative', async () => {
+      await expect(invoiceService.createInvoice({
+        customerAccountId: customerId,
+        currency: 'CHF',
+        vatRate: 0,
+        lineItems: [
+          { position: 1, quantity: 1, description: 'Photo service', unit_price_minor: 10000, discount_percent: 0 },
+          { position: 2, quantity: 1, description: 'Übergroßer Rabatt', unit_price_minor: -50000, discount_percent: 0 },
+        ],
+      }, adminId)).rejects.toMatchObject({
+        code: 'INVOICE_TOTAL_NEGATIVE',
+        statusCode: 400,
+      });
+    });
+  });
+});

backend/__tests__/integration/emailTemplateBoot.test.js

@@ -0,0 +1,98 @@
+/**
+ * Boot-time email-template self-heal:
+ *   1. Seeds the CRM / contract / event-reminder templates on an
+ *      install that's never had them before.
+ *   2. Recovers email_queue rows that previously exhausted their
+ *      retries because their template was missing.
+ *
+ * The failure that triggered this fix (2026-05-27) had Ralf's beta
+ * box failing every `quote_sent` / `invoice_sent` send for ~14h
+ * because crmEmailTemplates.ensureCrmEmailTemplatesSeeded was
+ * defined but never called. After 3 retries the rows sat in
+ * status='pending' forever; nothing in the admin UI signalled the
+ * problem. Both halves of that regression are covered here.
+ */
+
+const { bootCrmDb } = require('./helpers/crmDb');
+
+describe('email template self-heal at boot', () => {
+  let db;
+  let cleanup;
+
+  beforeAll(async () => {
+    ({ db, cleanup } = await bootCrmDb());
+  }, 120000);
+
+  afterAll(async () => {
+    if (cleanup) await cleanup();
+  });
+
+  it('seeds crm/contract/event-reminder templates and recovers stuck queue rows', async () => {
+    // Sanity: a fresh CRM-migrated DB does NOT carry CRM templates —
+    // 107_crm_consolidated documents the deliberate split (templates
+    // are self-healed at runtime, not inserted by the migration).
+    const before = await db('email_templates')
+      .whereIn('template_key', ['quote_sent', 'invoice_sent', 'storno_issued'])
+      .pluck('template_key');
+    expect(before).toEqual([]);
+
+    // Seed a stuck queue row that mirrors what we found on Ralf's box:
+    // quote_sent send attempted 3 times, each time failed because the
+    // template didn't exist, queue processor gave up.
+    const queueRowIds = await db('email_queue').insert({
+      recipient_email: 'customer@example.com',
+      email_type: 'quote_sent',
+      email_data: JSON.stringify({ quote_number: 'Q-2026-0001' }),
+      status: 'pending',
+      retry_count: 3,
+      error_message: "Email template 'quote_sent' not found",
+      created_at: new Date(),
+    }).returning('id');
+    const queueRowId = typeof queueRowIds[0] === 'object' ? queueRowIds[0].id : queueRowIds[0];
+
+    // Also seed an UNRELATED stuck row (different template, NOT one
+    // we're going to insert) to confirm the recovery is targeted —
+    // it must not blanket-reset every retry-exhausted row.
+    const unrelatedIds = await db('email_queue').insert({
+      recipient_email: 'someone@example.com',
+      email_type: 'some_other_template',
+      email_data: JSON.stringify({}),
+      status: 'pending',
+      retry_count: 3,
+      error_message: 'SMTP timeout',
+      created_at: new Date(),
+    }).returning('id');
+    const unrelatedId = typeof unrelatedIds[0] === 'object' ? unrelatedIds[0].id : unrelatedIds[0];
+
+    // The seeders use module-level caches (`_seeded = true`). When
+    // jest runs this test in isolation that cache starts fresh; in
+    // the full suite no other test currently calls these seeders, so
+    // the first call here also runs the real work. Reset the cache
+    // defensively in case a future test changes that.
+    jest.resetModules();
+    const { seedEmailTemplatesAndRecoverQueue } = require('../../src/services/_emailTemplateBoot');
+
+    const result = await seedEmailTemplatesAndRecoverQueue(db, null);
+
+    // Templates landed.
+    expect(result.seeded).toEqual(expect.arrayContaining([
+      'quote_sent', 'invoice_sent', 'storno_issued',
+    ]));
+    const after = await db('email_templates')
+      .whereIn('template_key', ['quote_sent', 'invoice_sent', 'storno_issued'])
+      .pluck('template_key');
+    expect(after.sort()).toEqual(['invoice_sent', 'quote_sent', 'storno_issued']);
+
+    // Stuck quote_sent row was recovered.
+    expect(result.recovered).toBeGreaterThanOrEqual(1);
+    const recoveredRow = await db('email_queue').where({ id: queueRowId }).first();
+    expect(recoveredRow.retry_count).toBe(0);
+    expect(recoveredRow.error_message).toBeNull();
+    expect(recoveredRow.status).toBe('pending'); // ready for the next tick
+
+    // Unrelated stuck row was NOT touched.
+    const unrelatedRow = await db('email_queue').where({ id: unrelatedId }).first();
+    expect(unrelatedRow.retry_count).toBe(3);
+    expect(unrelatedRow.error_message).toBe('SMTP timeout');
+  });
+});

backend/__tests__/integration/eventTypeRename.test.js

@@ -0,0 +1,64 @@
+/**
+ * Renaming an event type's slug_prefix must CASCADE to everything keyed on the
+ * old slug, so a rename behaves like a rename rather than silently detaching
+ * existing events/quotes and orphaning the per-type pre-event reminder template.
+ */
+const { bootCrmDb, seedMinimal } = require('./helpers/crmDb');
+
+// bootCrmDb runs the full core-migration set in beforeAll.
+jest.setTimeout(120000);
+
+describe('event type slug rename cascade', () => {
+  let db;
+  let cleanup;
+  let customerId;
+  let eventTypeService;
+
+  beforeAll(async () => {
+    ({ db, cleanup } = await bootCrmDb());
+    ({ customerId } = await seedMinimal(db));
+    eventTypeService = require('../../src/services/eventTypeService');
+  }, 120000);
+
+  afterAll(async () => { if (cleanup) await cleanup(); });
+
+  it('re-points events + quotes + the reminder template from old slug to new', async () => {
+    // A non-system event type with slug 'party'.
+    const [typeId] = await db('event_types').insert({ name: 'Party', slug_prefix: 'party', is_active: true });
+
+    // An authored per-type reminder template + an event + a quote, all on 'party'.
+    await db('email_templates').insert({ template_key: 'event_reminder_party', subject_en: 'Party reminder' });
+    await db('events').insert({
+      event_type: 'party', password_hash: 'x', expires_at: new Date(Date.now() + 9e9).toISOString(),
+      is_active: true, is_archived: false, slug: 'party-ev', share_link: 'party-ev',
+      event_name: 'A party', event_date: '2026-09-01',
+    });
+    await db('quotes').insert({
+      quote_number: 'Q-PARTY-1', customer_account_id: customerId, issue_date: '2026-01-01', event_type: 'party',
+    });
+
+    // Rename the slug.
+    await eventTypeService.updateEventType(typeId, { slug_prefix: 'concert' });
+
+    // Event + quote follow the rename.
+    expect((await db('events').where({ slug: 'party-ev' }).first()).event_type).toBe('concert');
+    expect((await db('quotes').where({ quote_number: 'Q-PARTY-1' }).first()).event_type).toBe('concert');
+    // The authored reminder template moved (subject/body preserved), old key gone.
+    expect(await db('email_templates').where({ template_key: 'event_reminder_party' }).first()).toBeUndefined();
+    const moved = await db('email_templates').where({ template_key: 'event_reminder_concert' }).first();
+    expect(moved).toBeTruthy();
+    expect(moved.subject_en).toBe('Party reminder');
+  });
+
+  it('does not clobber an existing template for the new slug', async () => {
+    const [typeId] = await db('event_types').insert({ name: 'Gala', slug_prefix: 'gala', is_active: true });
+    await db('email_templates').insert({ template_key: 'event_reminder_gala', subject_en: 'old gala' });
+    await db('email_templates').insert({ template_key: 'event_reminder_soiree', subject_en: 'existing soiree' });
+
+    await eventTypeService.updateEventType(typeId, { slug_prefix: 'soiree' });
+
+    // Target already existed → left intact; source not force-merged over it.
+    expect((await db('email_templates').where({ template_key: 'event_reminder_soiree' }).first()).subject_en)
+      .toBe('existing soiree');
+  });
+});

backend/__tests__/integration/externalImportCaptureDate.test.js

@@ -0,0 +1,161 @@
+/**
+ * External imports must record captured_at (#1172).
+ *
+ * Managed uploads get it from photoProcessor, which external media never goes
+ * through — so every externally imported photo carried captured_at NULL, and
+ * the gallery's "Date Taken" sort fell back to uploaded_at through its
+ * COALESCE. On a library imported in two batches that ordered a 12-day trip by
+ * which folder was imported first: the reporter's first two days landed at
+ * positions 4204-5296 of 5555.
+ *
+ * Driven through the real route against real files carrying real EXIF, because
+ * the whole question is whether the import reads the file it already has open.
+ */
+
+const fs = require('fs');
+const path = require('path');
+const os = require('os');
+const express = require('express');
+const request = require('supertest');
+const sharp = require('sharp');
+
+describe('external import capture dates (#1172)', () => {
+  let tmpDir; let db; let app; let mediaRoot;
+
+  /**
+   * A real JPEG carrying DateTimeOriginal.
+   *
+   * IFD2, not IFD0 — DateTimeOriginal lives in the Exif IFD, and exifr does not
+   * see it anywhere else (IFD0 takes plain DateTime, which surfaces as
+   * ModifyDate instead).
+   */
+  const writeJpegWithExif = async (rel, iso) => {
+    const full = path.join(mediaRoot, rel);
+    await fs.promises.mkdir(path.dirname(full), { recursive: true });
+    const d = new Date(iso);
+    const pad = (n) => String(n).padStart(2, '0');
+    const exifDate = `${d.getUTCFullYear()}:${pad(d.getUTCMonth() + 1)}:${pad(d.getUTCDate())} `
+      + `${pad(d.getUTCHours())}:${pad(d.getUTCMinutes())}:${pad(d.getUTCSeconds())}`;
+    await sharp({ create: { width: 60, height: 40, channels: 3, background: { r: 10, g: 20, b: 30 } } })
+      .withExif({ IFD2: { DateTimeOriginal: exifDate } })
+      .jpeg()
+      .toFile(full);
+    return full;
+  };
+
+  const writeJpegNoExif = async (rel) => {
+    const full = path.join(mediaRoot, rel);
+    await fs.promises.mkdir(path.dirname(full), { recursive: true });
+    await sharp({ create: { width: 60, height: 40, channels: 3, background: { r: 200, g: 10, b: 10 } } })
+      .jpeg().toFile(full);
+  };
+
+  beforeAll(async () => {
+    tmpDir = await fs.promises.mkdtemp(path.join(os.tmpdir(), 'picpeak-capdate-'));
+    mediaRoot = path.join(tmpDir, 'media');
+    await fs.promises.mkdir(mediaRoot, { recursive: true });
+
+    process.env.NODE_ENV = 'test';
+    process.env.TEST_DATABASE_PATH = path.join(tmpDir, 'data', 'db.sqlite');
+    await fs.promises.mkdir(path.dirname(process.env.TEST_DATABASE_PATH), { recursive: true });
+    process.env.STORAGE_PATH = path.join(tmpDir, 'storage');
+    process.env.EXTERNAL_MEDIA_ROOT = mediaRoot;
+    process.env.JWT_SECRET = process.env.JWT_SECRET || 'capdate-secret';
+
+    jest.resetModules();
+    jest.doMock('../../src/middleware/auth', () => ({
+      adminAuth: (req, _res, next) => { req.admin = { id: 1, username: 'tester', roleName: 'admin' }; next(); },
+    }));
+    jest.doMock('../../src/middleware/permissions', () => ({
+      requirePermission: () => (_req, _res, next) => next(),
+    }));
+    jest.doMock('../../src/middleware/ownership', () => ({
+      requireEventOwnership: (_req, _res, next) => next(),
+    }));
+    jest.doMock('../../src/services/imageProcessor', () => {
+      const actual = jest.requireActual('../../src/services/imageProcessor');
+      return { ...actual, generateThumbnail: jest.fn(async () => 'thumbnails/mock.jpg'), ensureThumbnail: jest.fn() };
+    });
+    jest.doMock('../../src/utils/logger', () => ({
+      debug: jest.fn(), info: jest.fn(), warn: jest.fn(), error: jest.fn(),
+    }));
+
+    ({ db } = await require('./helpers/crmDb').bootCrmDb());
+
+    app = express();
+    app.use(express.json());
+    app.use('/api/admin/external-media', require('../../src/routes/adminExternalMedia'));
+  }, 180000);
+
+  afterAll(async () => {
+    if (db) await db.destroy?.();
+    await fs.promises.rm(tmpDir, { recursive: true, force: true }).catch(() => {});
+  });
+
+  async function seedEvent() {
+    await db('photos').del();
+    await db('events').del();
+    await fs.promises.rm(mediaRoot, { recursive: true, force: true });
+    await fs.promises.mkdir(mediaRoot, { recursive: true });
+    const [e] = await db('events').insert({
+      slug: `capdate-${Math.random().toString(36).slice(2, 8)}`,
+      event_type: 'wedding', event_name: 'capdate', event_date: '2026-01-01',
+      host_email: 'h@example.com', admin_email: 'a@example.com', password_hash: 'x',
+      share_link: `capdate-${Math.random()}`, expires_at: new Date().toISOString(),
+      source_mode: 'reference',
+    }).returning('id');
+    return typeof e === 'object' ? e.id : e;
+  }
+
+  const runImport = (eventId, external_path) => request(app)
+    .post(`/api/admin/external-media/events/${eventId}/import-external`)
+    .send({ external_path, recursive: true });
+
+  it('records the EXIF capture date on import', async () => {
+    const eventId = await seedEvent();
+    await writeJpegWithExif('trip/a.jpg', '2026-06-01T09:45:03Z');
+
+    await runImport(eventId, 'trip');
+
+    const photo = await db('photos').where({ event_id: eventId }).first();
+    expect(photo.captured_at).toBeTruthy();
+    // NOT asserted as an absolute instant. EXIF carries a naive wall-clock
+    // time and exifr resolves it against the HOST timezone, so the stored UTC
+    // value differs between a CEST developer machine and a UTC runner. What
+    // this fix is about is that the field is populated and orders correctly;
+    // that captured_at is not a true instant is a separate, pre-existing
+    // problem shared with managed uploads (#1172's own footnote).
+    expect(new Date(photo.captured_at).getUTCFullYear()).toBe(2026);
+    expect(new Date(photo.captured_at).getUTCMonth()).toBe(5); // June
+  });
+
+  it('imports a photo with no EXIF date rather than failing it', async () => {
+    // Plenty of sources carry none; that must stay an import, not an error.
+    const eventId = await seedEvent();
+    await writeJpegNoExif('trip/plain.jpg');
+
+    const res = await runImport(eventId, 'trip');
+
+    expect(res.body.imported).toBe(1);
+    const photo = await db('photos').where({ event_id: eventId }).first();
+    expect(photo.captured_at).toBeNull();
+  });
+
+  it('orders a two-batch import by capture time, not by batch', async () => {
+    // The reported shape: the FIRST days of the trip imported second. Sorting
+    // on COALESCE(captured_at, uploaded_at) put them after the last days,
+    // because uploaded_at is the import timestamp.
+    const eventId = await seedEvent();
+    await writeJpegWithExif('late/day12.jpg', '2026-06-12T10:00:00Z');
+    await runImport(eventId, 'late');
+    await writeJpegWithExif('early/day01.jpg', '2026-06-01T10:00:00Z');
+    await runImport(eventId, 'early');
+
+    const rows = await db('photos')
+      .where({ event_id: eventId })
+      .orderByRaw('COALESCE(captured_at, uploaded_at) asc')
+      .select('filename');
+
+    expect(rows.map((r) => r.filename)).toEqual(['day01.jpg', 'day12.jpg']);
+  });
+});

backend/__tests__/integration/externalImportConcurrency.test.js

@@ -0,0 +1,205 @@
+/**
+ * Two overlapping external imports insert every file twice (#1162).
+ *
+ * The route checked for an existing external_relpath and then inserted, with
+ * an fs.stat and a `sharp().metadata()` read sitting in between. A reporter
+ * double-clicked a slow import of a 6012-file tree and got 8004 rows.
+ *
+ * Both halves of the fix are driven here through the real route:
+ *
+ *   - the in-flight guard, which turns the second click into a 409 instead of
+ *     a second full walk of the tree;
+ *   - convergence when the guard cannot help (another replica, another
+ *     process), which is the unique index from migration 186 firing and the
+ *     loop counting a skip rather than dying or duplicating.
+ *
+ * The second is exercised by inserting a competing row from inside the mocked
+ * `sharp().metadata()` call — literally inside the window the bug lived in.
+ */
+
+const fs = require('fs');
+const path = require('path');
+const os = require('os');
+const express = require('express');
+const request = require('supertest');
+
+describe('concurrent external imports (#1162)', () => {
+  let tmpDir; let db; let app; let mediaRoot;
+  // When set, the mocked sharp metadata read inserts this row first — the
+  // other run winning the race between our SELECT and our INSERT.
+  let stealDuringMetadata = null;
+  let thumbnailDelayMs = 0;
+
+  beforeAll(async () => {
+    tmpDir = await fs.promises.mkdtemp(path.join(os.tmpdir(), 'picpeak-extdup-'));
+    mediaRoot = path.join(tmpDir, 'media');
+    await fs.promises.mkdir(path.join(mediaRoot, 'nas', 'individual'), { recursive: true });
+    for (const name of ['a.jpg', 'b.jpg', 'c.jpg']) {
+      await fs.promises.writeFile(path.join(mediaRoot, 'nas', 'individual', name), 'not-a-real-jpeg');
+    }
+
+    process.env.NODE_ENV = 'test';
+    process.env.TEST_DATABASE_PATH = path.join(tmpDir, 'data', 'db.sqlite');
+    await fs.promises.mkdir(path.dirname(process.env.TEST_DATABASE_PATH), { recursive: true });
+    process.env.STORAGE_PATH = path.join(tmpDir, 'storage');
+    process.env.EXTERNAL_MEDIA_ROOT = mediaRoot;
+    process.env.JWT_SECRET = process.env.JWT_SECRET || 'extdup-secret';
+
+    jest.resetModules();
+
+    jest.doMock('../../src/middleware/auth', () => ({
+      adminAuth: (req, _res, next) => { req.admin = { id: 1, username: 'tester', roleName: 'admin' }; next(); },
+    }));
+    jest.doMock('../../src/middleware/permissions', () => ({
+      requirePermission: () => (_req, _res, next) => next(),
+    }));
+    jest.doMock('../../src/middleware/ownership', () => ({
+      requireEventOwnership: (_req, _res, next) => next(),
+    }));
+
+    // The window. In production this is a real decode of a NAS-hosted file —
+    // hundreds of milliseconds during which the row we just proved absent can
+    // appear. Standing in for the other run here makes that deterministic.
+    jest.doMock('sharp', () => () => ({
+      metadata: async () => {
+        if (stealDuringMetadata) {
+          const { db: liveDb } = require('../../src/database/db');
+          await liveDb('photos').insert(stealDuringMetadata);
+          stealDuringMetadata = null;
+        }
+        return { width: 100, height: 200 };
+      },
+    }));
+
+    jest.doMock('../../src/services/imageProcessor', () => ({
+      generateThumbnail: jest.fn(async () => {
+        if (thumbnailDelayMs) await new Promise((r) => setTimeout(r, thumbnailDelayMs));
+        return 'thumbnails/mock.jpg';
+      }),
+      ensureThumbnail: jest.fn(),
+    }));
+
+    jest.doMock('../../src/utils/logger', () => ({
+      debug: jest.fn(), info: jest.fn(), warn: jest.fn(), error: jest.fn(),
+    }));
+
+    ({ db } = await require('./helpers/crmDb').bootCrmDb());
+
+    app = express();
+    app.use(express.json());
+    app.use('/api/admin/external-media', require('../../src/routes/adminExternalMedia'));
+  }, 180000);
+
+  afterAll(async () => {
+    if (db) await db.destroy?.();
+    await fs.promises.rm(tmpDir, { recursive: true, force: true }).catch(() => {});
+  });
+
+  async function seedEvent() {
+    await db('photos').del();
+    await db('events').del();
+    stealDuringMetadata = null;
+    thumbnailDelayMs = 0;
+    const [e] = await db('events').insert({
+      slug: `extdup-${Math.random().toString(36).slice(2, 8)}`,
+      event_type: 'wedding',
+      event_name: 'extdup',
+      event_date: '2026-01-01',
+      host_email: 'h@example.com',
+      admin_email: 'a@example.com',
+      password_hash: 'x',
+      share_link: `extdup-${Math.random()}`,
+      expires_at: new Date().toISOString(),
+      source_mode: 'reference',
+    }).returning('id');
+    return typeof e === 'object' ? e.id : e;
+  }
+
+  const runImport = (eventId) => request(app)
+    .post(`/api/admin/external-media/events/${eventId}/import-external`)
+    .send({ external_path: 'nas', recursive: true });
+
+  async function relpathCounts(eventId) {
+    const rows = await db('photos').where({ event_id: eventId }).select('external_relpath');
+    const counts = new Map();
+    for (const r of rows) counts.set(r.external_relpath, (counts.get(r.external_relpath) || 0) + 1);
+    return counts;
+  }
+
+  it('rejects a second import while the first is still running', async () => {
+    const eventId = await seedEvent();
+    // Enough to keep the first request inside its loop while the second
+    // arrives — the "slow import looks hung, so I clicked again" case.
+    thumbnailDelayMs = 20;
+
+    const [first, second] = await Promise.all([runImport(eventId), runImport(eventId)]);
+
+    const statuses = [first.status, second.status].sort();
+    expect(statuses).toEqual([200, 409]);
+    const rejected = first.status === 409 ? first : second;
+    expect(rejected.body.error).toMatch(/already running/i);
+  });
+
+  it('leaves exactly one row per file after both runs', async () => {
+    const eventId = await seedEvent();
+    thumbnailDelayMs = 20;
+
+    await Promise.all([runImport(eventId), runImport(eventId)]);
+
+    const counts = await relpathCounts(eventId);
+    expect(counts.size).toBe(3);
+    expect([...counts.values()]).toEqual([1, 1, 1]);
+  });
+
+  it('releases the event once the import finishes, so a re-import still works', async () => {
+    const eventId = await seedEvent();
+
+    expect((await runImport(eventId)).status).toBe(200);
+    // Not 409 — the guard is per run, not a permanent lock on the event.
+    const second = await runImport(eventId);
+    expect(second.status).toBe(200);
+    expect(second.body.imported).toBe(0);
+    expect(second.body.skipped).toBe(3);
+  });
+
+  it('converges when another writer wins the race mid-file', async () => {
+    // The guard is in-process, so it cannot see a second replica. This is what
+    // the unique index is for: the insert bounces, and the file is counted as
+    // skipped rather than duplicated or lost to a 500.
+    const eventId = await seedEvent();
+    stealDuringMetadata = {
+      event_id: eventId,
+      filename: 'a.jpg',
+      path: 'x/a.jpg',
+      type: 'individual',
+      source_origin: 'external',
+      external_relpath: path.join('nas', 'individual', 'a.jpg'),
+    };
+
+    const res = await runImport(eventId);
+
+    expect(res.status).toBe(200);
+    const counts = await relpathCounts(eventId);
+    expect(counts.get(path.join('nas', 'individual', 'a.jpg'))).toBe(1);
+    // Two imported by us, one lost to the other writer and reported honestly.
+    expect(res.body.imported).toBe(2);
+    expect(res.body.skipped).toBe(1);
+  });
+
+  it('does not let one contended file abort the rest of the import', async () => {
+    const eventId = await seedEvent();
+    stealDuringMetadata = {
+      event_id: eventId,
+      filename: 'a.jpg',
+      path: 'x/a.jpg',
+      type: 'individual',
+      source_origin: 'external',
+      external_relpath: path.join('nas', 'individual', 'a.jpg'),
+    };
+
+    await runImport(eventId);
+
+    // All three files present — the contended one via the other writer's row.
+    expect((await relpathCounts(eventId)).size).toBe(3);
+  });
+});

backend/__tests__/integration/externalImportSecondFolder.test.js

@@ -0,0 +1,175 @@
+/**
+ * Importing a second folder must not move the photos already in the event (#1163).
+ *
+ * events.external_path is overwritten by every import, and external_relpath
+ * used to be stored relative to it — so a second import silently rebased every
+ * existing row onto the new folder. The reporter had 7547 of 8004 originals
+ * pointing at files that do not exist, and nothing said so: thumbnails are
+ * written to local storage during the import while the base path is still
+ * correct, so the grid carries on rendering.
+ *
+ * Driven through the real route and the real resolver, against a real
+ * directory tree — the failure is entirely about whether a file is where the
+ * app looks for it.
+ */
+
+const fs = require('fs');
+const path = require('path');
+const os = require('os');
+const express = require('express');
+const request = require('supertest');
+
+describe('a second external import (#1163)', () => {
+  let tmpDir; let db; let app; let mediaRoot; let resolvePhotoFilePath;
+
+  const touch = async (rel) => {
+    const full = path.join(mediaRoot, rel);
+    await fs.promises.mkdir(path.dirname(full), { recursive: true });
+    await fs.promises.writeFile(full, 'not-a-real-jpeg');
+  };
+
+  beforeAll(async () => {
+    tmpDir = await fs.promises.mkdtemp(path.join(os.tmpdir(), 'picpeak-ext2nd-'));
+    mediaRoot = path.join(tmpDir, 'media');
+    await fs.promises.mkdir(mediaRoot, { recursive: true });
+
+    process.env.NODE_ENV = 'test';
+    process.env.TEST_DATABASE_PATH = path.join(tmpDir, 'data', 'db.sqlite');
+    await fs.promises.mkdir(path.dirname(process.env.TEST_DATABASE_PATH), { recursive: true });
+    process.env.STORAGE_PATH = path.join(tmpDir, 'storage');
+    process.env.EXTERNAL_MEDIA_ROOT = mediaRoot;
+    process.env.JWT_SECRET = process.env.JWT_SECRET || 'ext2nd-secret';
+
+    jest.resetModules();
+
+    jest.doMock('../../src/middleware/auth', () => ({
+      adminAuth: (req, _res, next) => { req.admin = { id: 1, username: 'tester', roleName: 'admin' }; next(); },
+    }));
+    jest.doMock('../../src/middleware/permissions', () => ({
+      requirePermission: () => (_req, _res, next) => next(),
+    }));
+    jest.doMock('../../src/middleware/ownership', () => ({
+      requireEventOwnership: (_req, _res, next) => next(),
+    }));
+    jest.doMock('sharp', () => () => ({ metadata: async () => ({ width: 100, height: 200 }) }));
+    jest.doMock('../../src/services/imageProcessor', () => ({
+      generateThumbnail: jest.fn(async () => 'thumbnails/mock.jpg'),
+      ensureThumbnail: jest.fn(),
+    }));
+    jest.doMock('../../src/utils/logger', () => ({
+      debug: jest.fn(), info: jest.fn(), warn: jest.fn(), error: jest.fn(),
+    }));
+
+    ({ db } = await require('./helpers/crmDb').bootCrmDb());
+    ({ resolvePhotoFilePath } = require('../../src/services/photoResolver'));
+
+    app = express();
+    app.use(express.json());
+    app.use('/api/admin/external-media', require('../../src/routes/adminExternalMedia'));
+  }, 180000);
+
+  afterAll(async () => {
+    if (db) await db.destroy?.();
+    await fs.promises.rm(tmpDir, { recursive: true, force: true }).catch(() => {});
+  });
+
+  async function seedEvent() {
+    await db('photos').del();
+    await db('events').del();
+    await fs.promises.rm(mediaRoot, { recursive: true, force: true });
+    await fs.promises.mkdir(mediaRoot, { recursive: true });
+    const [e] = await db('events').insert({
+      slug: `ext2nd-${Math.random().toString(36).slice(2, 8)}`,
+      event_type: 'wedding',
+      event_name: 'ext2nd',
+      event_date: '2026-01-01',
+      host_email: 'h@example.com',
+      admin_email: 'a@example.com',
+      password_hash: 'x',
+      share_link: `ext2nd-${Math.random()}`,
+      expires_at: new Date().toISOString(),
+      source_mode: 'reference',
+    }).returning('id');
+    return typeof e === 'object' ? e.id : e;
+  }
+
+  const runImport = (eventId, external_path) => request(app)
+    .post(`/api/admin/external-media/events/${eventId}/import-external`)
+    .send({ external_path, recursive: true });
+
+  /** Where the app would go looking for this photo's original, right now. */
+  async function resolved(eventId, filename) {
+    const event = await db('events').where({ id: eventId }).first();
+    const photo = await db('photos').where({ event_id: eventId, filename }).first();
+    return resolvePhotoFilePath(event, photo);
+  }
+
+  it('stores paths relative to the media root, not to the imported folder', async () => {
+    const eventId = await seedEvent();
+    await touch('Trip/Leknes/old.jpg');
+
+    await runImport(eventId, 'Trip');
+
+    const photo = await db('photos').where({ event_id: eventId }).first();
+    expect(photo.external_relpath).toBe(path.join('Trip', 'Leknes', 'old.jpg'));
+  });
+
+  it('leaves the first folder’s originals reachable after a second import', async () => {
+    const eventId = await seedEvent();
+    await touch('Trip/Leknes/old.jpg');
+    await touch('Trip/Sub/new.jpg');
+
+    await runImport(eventId, 'Trip');
+    const before = await resolved(eventId, 'old.jpg');
+    await runImport(eventId, 'Trip/Sub');
+    const after = await resolved(eventId, 'old.jpg');
+
+    // The regression: `after` used to be <root>/Trip/Sub/Leknes/old.jpg.
+    expect(after).toBe(before);
+    expect(fs.existsSync(after)).toBe(true);
+  });
+
+  it('every original in the event is still on disk afterwards', async () => {
+    const eventId = await seedEvent();
+    await touch('Trip/Leknes/a.jpg');
+    await touch('Trip/Leknes/b.jpg');
+    await touch('Trip/Sub/c.jpg');
+
+    await runImport(eventId, 'Trip');
+    await runImport(eventId, 'Trip/Sub');
+
+    const event = await db('events').where({ id: eventId }).first();
+    const photos = await db('photos').where({ event_id: eventId });
+    expect(photos).toHaveLength(3);
+    for (const photo of photos) {
+      expect(fs.existsSync(resolvePhotoFilePath(event, photo))).toBe(true);
+    }
+  });
+
+  it('does not re-insert a file the first import already took', async () => {
+    // The dedupe check compares stored paths, so it has to be comparing the
+    // same shape the insert writes.
+    const eventId = await seedEvent();
+    await touch('Trip/Sub/c.jpg');
+
+    await runImport(eventId, 'Trip');
+    const second = await runImport(eventId, 'Trip/Sub');
+
+    expect(second.body.imported).toBe(0);
+    expect(second.body.skipped).toBe(1);
+    expect(await db('photos').where({ event_id: eventId }).count('* as c').first()).toEqual({ c: 1 });
+  });
+
+  it('resolves a subfolder that repeats its parent’s name', async () => {
+    // The old resolver stripped the relpath's first segment when it matched the
+    // base path's last one, which broke exactly this layout.
+    const eventId = await seedEvent();
+    await touch('Trip/Trip/x.jpg');
+
+    await runImport(eventId, 'Trip');
+
+    const event = await db('events').where({ id: eventId }).first();
+    const photo = await db('photos').where({ event_id: eventId }).first();
+    expect(resolvePhotoFilePath(event, photo)).toBe(path.join(mediaRoot, 'Trip', 'Trip', 'x.jpg'));
+  });
+});

backend/__tests__/integration/externalRelpathFoldPg.test.js

@@ -0,0 +1,121 @@
+/**
+ * PostgreSQL integration test for the external-path fold (#1163).
+ *
+ * Gated the same way as picpeakRestorePg: runs only when PICPEAK_PG_TEST_URL
+ * points at a throwaway Postgres DB, e.g.
+ *   PICPEAK_PG_TEST_URL="postgres://picpeak:picpeak_secure_pass_2024@127.0.0.1:7102/picpeak_fold_test" \
+ *     npx jest __tests__/integration/externalRelpathFoldPg.test.js
+ *
+ * This exists because of a defect SQLite could not have caught. The two-pass
+ * rewrite parks each row on a temporary value, and that value was first written
+ * with a leading NUL. SQLite stores NUL in TEXT without complaint; Postgres
+ * rejects it outright ("invalid byte sequence for encoding UTF8"), so migration
+ * 187 would have rolled back on exactly the installs needing the repair — and
+ * only on the engine most of them run.
+ *
+ * The staging value is therefore an engine-level contract, not an
+ * implementation detail, and it is pinned here on the engine that constrains it.
+ */
+
+const knex = require('knex');
+const fs = require('fs');
+const os = require('os');
+const path = require('path');
+
+const PG_URL = process.env.PICPEAK_PG_TEST_URL;
+const maybe = PG_URL ? describe : describe.skip;
+
+maybe('external relpath fold on Postgres', () => {
+  let pgDb; let mediaRoot; let fold;
+
+  const touch = async (rel, bytes) => {
+    const full = path.join(mediaRoot, rel);
+    await fs.promises.mkdir(path.dirname(full), { recursive: true });
+    await fs.promises.writeFile(full, Buffer.alloc(bytes));
+    return bytes;
+  };
+
+  beforeAll(async () => {
+    mediaRoot = await fs.promises.mkdtemp(path.join(os.tmpdir(), 'picpeak-foldpg-'));
+    process.env.EXTERNAL_MEDIA_ROOT = mediaRoot;
+    jest.resetModules();
+    ({ foldExternalRelpaths: fold } = require('../../src/services/externalRelpathFold'));
+
+    pgDb = knex({ client: 'pg', connection: PG_URL });
+  }, 60000);
+
+  afterAll(async () => {
+    if (pgDb) await pgDb.destroy();
+    await fs.promises.rm(mediaRoot, { recursive: true, force: true }).catch(() => {});
+    delete process.env.EXTERNAL_MEDIA_ROOT;
+  });
+
+  beforeEach(async () => {
+    await pgDb.raw('DROP TABLE IF EXISTS photos, events, app_settings CASCADE');
+    await pgDb.schema.createTable('events', (t) => {
+      t.increments('id');
+      t.text('external_path');
+    });
+    await pgDb.schema.createTable('photos', (t) => {
+      t.increments('id');
+      t.integer('event_id');
+      t.text('external_relpath');
+      t.bigInteger('size_bytes');
+      t.string('source_origin').defaultTo('managed');
+    });
+    await pgDb.schema.createTable('app_settings', (t) => {
+      t.increments('id');
+      t.string('setting_key');
+      t.text('setting_value');
+      t.string('setting_type');
+      t.string('updated_at');
+    });
+    await fs.promises.rm(mediaRoot, { recursive: true, force: true });
+    await fs.promises.mkdir(mediaRoot, { recursive: true });
+  });
+
+  const relpaths = async () =>
+    (await pgDb('photos').orderBy('id').select('external_relpath')).map((r) => r.external_relpath);
+
+  it('completes the two-pass repair that a NUL staging value would abort', async () => {
+    // The exact shape that forces staging: `photo.jpg` repairs up to
+    // `Trip/photo.jpg`, while the row already holding `Trip/photo.jpg` folds
+    // deeper. Every final value is distinct, but a final value equals another
+    // row's current one, so the rewrite has to park first.
+    const a = await touch('Trip/photo.jpg', 11);
+    const b = await touch('Trip/Sub/Trip/photo.jpg', 22);
+    await pgDb('events').insert({ id: 1, external_path: 'Trip/Sub' });
+    await pgDb('photos').insert([
+      { event_id: 1, external_relpath: 'photo.jpg', size_bytes: a, source_origin: 'external' },
+      { event_id: 1, external_relpath: 'Trip/photo.jpg', size_bytes: b, source_origin: 'external' },
+    ]);
+
+    await fold(pgDb);
+
+    expect(await relpaths()).toEqual(['Trip/photo.jpg', 'Trip/Sub/Trip/photo.jpg']);
+  });
+
+  it('leaves no staging value behind', async () => {
+    await touch('Trip/a.jpg', 8);
+    await pgDb('events').insert({ id: 1, external_path: 'Trip' });
+    await pgDb('photos').insert({ event_id: 1, external_relpath: 'a.jpg', size_bytes: 8, source_origin: 'external' });
+
+    await fold(pgDb);
+
+    const rows = await relpaths();
+    expect(rows).toEqual(['Trip/a.jpg']);
+    expect(rows.some((r) => r.includes('staging'))).toBe(false);
+  });
+
+  it('folds and marks in one transaction', async () => {
+    await touch('Trip/a.jpg', 8);
+    await pgDb('events').insert({ id: 1, external_path: 'Trip' });
+    await pgDb('photos').insert({ event_id: 1, external_relpath: 'a.jpg', size_bytes: 8, source_origin: 'external' });
+
+    await fold(pgDb);
+    // Second run is a no-op: the marker committed with the rewrites.
+    await fold(pgDb);
+
+    expect(await relpaths()).toEqual(['Trip/a.jpg']);
+  });
+});

backend/__tests__/integration/galleryFilterFeedbackVisibility.test.js

@@ -0,0 +1,216 @@
+/**
+ * Guest filters must respect show_feedback_to_guests (#1044 follow-up).
+ *
+ * Every filter token on /photos is an OR of two halves: what THIS viewer
+ * marked, and what ANYONE marked. The response fields built from the second
+ * half — like_count, comment_count — are all gated on
+ * show_feedback_to_guests. The FILTER was not.
+ *
+ * So with the setting off, the numbers were hidden but `?filter=liked` still
+ * returned exactly the photos other people had liked: the same information as
+ * a set instead of a count, one token at a time. These tests pin the gate on
+ * every token, and pin that the viewer's own half is never gated — filtering
+ * by what you yourself marked is yours to do regardless.
+ */
+
+const request = require('supertest');
+const express = require('express');
+const cookieParser = require('cookie-parser');
+const jwt = require('jsonwebtoken');
+
+const { bootCrmDb, seedMinimal } = require('./helpers/crmDb');
+
+process.env.JWT_SECRET = process.env.JWT_SECRET || 'filter-visibility-secret';
+
+const SLUG = 'filter-visibility-event';
+const ME = 'guest-me-identifier';
+const SOMEONE_ELSE = 'guest-other-identifier';
+
+describe('guest filters and show_feedback_to_guests (#1044)', () => {
+  let db;
+  let cleanup;
+  let app;
+  let eventId;
+  let mine;
+  let theirs;
+  let myGuestRowId;
+
+  const galleryToken = () => jwt.sign(
+    { eventId, eventSlug: SLUG, type: 'gallery' },
+    process.env.JWT_SECRET,
+    { expiresIn: '1h', issuer: 'picpeak-auth' }
+  );
+
+  const setVisibility = (visible) => db('event_feedback_settings')
+    .where({ event_id: eventId })
+    .update({ show_feedback_to_guests: visible });
+
+  // A real verified guest, which is how the viewer's own feedback is actually
+  // identified — NOT the `guest_id` query parameter the frontend invents.
+  const guestToken = () => jwt.sign(
+    { type: 'guest', guestId: myGuestRowId, eventId },
+    process.env.JWT_SECRET,
+    { expiresIn: '1h', issuer: 'picpeak-auth' }
+  );
+
+  const filter = async (token, { as = 'me', claimGuestId } = {}) => {
+    const req = request(app)
+      .get(`/api/gallery/${SLUG}/photos`)
+      .query({ filter: token, ...(claimGuestId ? { guest_id: claimGuestId } : {}) })
+      .set('Authorization', `Bearer ${galleryToken()}`);
+    if (as === 'me') req.set('x-guest-token', guestToken());
+    const res = await req;
+    expect(res.status).toBe(200);
+    const photos = Array.isArray(res.body) ? res.body : res.body.photos;
+    return (photos || []).map((p) => p.id).sort((a, b) => a - b);
+  };
+
+  beforeAll(async () => {
+    ({ db, cleanup } = await bootCrmDb());
+    await seedMinimal(db);
+
+    const inserted = await db('events').insert({
+      slug: SLUG,
+      event_type: 'wedding',
+      event_name: 'Filter Visibility',
+      event_date: '2026-08-01',
+      host_email: 'host@example.com',
+      admin_email: 'admin@example.com',
+      password_hash: 'x',
+      share_link: `/gallery/${SLUG}/share`,
+      share_token: 'filter-visibility-share',
+      expires_at: new Date(Date.now() + 7 * 24 * 3600 * 1000).toISOString(),
+      is_active: 1,
+      is_archived: 0,
+      is_draft: 0,
+      created_at: new Date().toISOString(),
+    }).returning('id');
+    eventId = inserted[0]?.id ?? inserted[0];
+
+    const addPhoto = async (name) => {
+      const p = await db('photos').insert({
+        event_id: eventId,
+        filename: name,
+        path: `events/filter/${name}`,
+        type: 'individual',
+        uploaded_at: new Date().toISOString(),
+      }).returning('id');
+      return p[0]?.id ?? p[0];
+    };
+    mine = await addPhoto('mine.jpg');
+    theirs = await addPhoto('theirs.jpg');
+
+    await db('event_feedback_settings').insert({
+      event_id: eventId,
+      feedback_enabled: true,
+      allow_likes: true,
+      allow_comments: true,
+      allow_ratings: true,
+      allow_favorites: true,
+      moderate_comments: false,
+      show_feedback_to_guests: true,
+    });
+
+    const guestRow = await db('gallery_guests').insert({
+      event_id: eventId,
+      name: 'Me',
+      identifier: ME,
+      created_at: new Date().toISOString(),
+      last_seen_at: new Date().toISOString(),
+      is_deleted: false,
+    }).returning('id');
+    myGuestRowId = guestRow[0]?.id ?? guestRow[0];
+
+    const feedback = (photoId, who, type, extra = {}) => db('photo_feedback').insert({
+      photo_id: photoId,
+      event_id: eventId,
+      guest_identifier: who,
+      // Submission links to the per-person guest row when one is present, and
+      // that is the column the viewer's own half resolves through.
+      guest_id: who === ME ? myGuestRowId : null,
+      feedback_type: type,
+      is_approved: true,
+      is_hidden: false,
+      created_at: new Date().toISOString(),
+      ...extra,
+    });
+
+    // Everything on `theirs` belongs to somebody else; `mine` is this viewer's.
+    await feedback(mine, ME, 'like');
+    await feedback(theirs, SOMEONE_ELSE, 'like');
+    await feedback(theirs, SOMEONE_ELSE, 'favorite');
+    await feedback(theirs, SOMEONE_ELSE, 'comment', { comment_text: 'lovely' });
+    await feedback(theirs, SOMEONE_ELSE, 'rating', { rating: 5 });
+
+    // The denormalized counters the aggregate half of the filter reads.
+    await db('photos').where('id', theirs).update({
+      like_count: 1, favorite_count: 1, comment_count: 1, average_rating: 5,
+    });
+    await db('photos').where('id', mine).update({ like_count: 1 });
+
+    app = express();
+    app.use(express.json());
+    app.use(cookieParser());
+    app.use('/api/gallery', require('../../src/routes/gallery'));
+  }, 120000);
+
+  afterAll(async () => {
+    if (cleanup) await cleanup();
+  });
+
+  describe('with feedback visible to guests', () => {
+    beforeAll(() => setVisibility(true));
+
+    it('shows other people\'s marks through every token, as before', async () => {
+      expect(await filter('liked')).toEqual([mine, theirs].sort((a, b) => a - b));
+      expect(await filter('favorited')).toEqual([theirs]);
+      expect(await filter('rated')).toEqual([theirs]);
+      expect(await filter('commented')).toEqual([theirs]);
+    });
+  });
+
+  describe('with feedback hidden from guests', () => {
+    beforeAll(() => setVisibility(false));
+
+    it('stops every token from selecting on other people\'s marks', async () => {
+      // `theirs` is the photo only other guests marked. It must not come back
+      // through any token — a filter that selects on hidden feedback reports
+      // that feedback just as surely as a count would.
+      expect(await filter('favorited')).toEqual([]);
+      expect(await filter('rated')).toEqual([]);
+      expect(await filter('commented')).toEqual([]);
+    });
+
+    it('still filters by what the viewer marked themselves', async () => {
+      // The viewer's own half is never gated: this is their own action, and
+      // hiding it would break "show me the ones I liked" for no privacy gain.
+      expect(await filter('liked')).toEqual([mine]);
+    });
+
+    it('drops the viewer\'s own feedback once an admin hides it', async () => {
+      // Moderation has to reach the filter too. getPhotoFeedback excludes
+      // hidden rows for the guest's OWN feedback, so a photo matching here
+      // would come back with nothing visible on it to explain why.
+      await db('photo_feedback')
+        .where({ photo_id: mine, guest_id: myGuestRowId, feedback_type: 'like' })
+        .update({ is_hidden: true });
+
+      expect(await filter('liked')).toEqual([]);
+
+      await db('photo_feedback')
+        .where({ photo_id: mine, guest_id: myGuestRowId, feedback_type: 'like' })
+        .update({ is_hidden: false });
+      expect(await filter('liked')).toEqual([mine]);
+    });
+
+    it('ignores a guest_id supplied by the caller', async () => {
+      // The own-half is resolved from the request identity. If it honoured the
+      // query string instead, anyone holding another guest's identifier could
+      // read that guest's hidden memberships one token at a time — straight
+      // back through the gate this file exists to pin.
+      expect(await filter('favorited', { claimGuestId: SOMEONE_ELSE })).toEqual([]);
+      // And an anonymous caller claiming to be me gets nothing of mine.
+      expect(await filter('liked', { as: 'anon', claimGuestId: ME })).toEqual([]);
+    });
+  });
+});

backend/__tests__/integration/galleryShortUrlRoute.test.js

@@ -0,0 +1,231 @@
+/**
+ * HTTP-level tests for the `/s/:shortSlug` public resolver (#699).
+ *
+ * Verifies the contract the public route is expected to honour:
+ *   - Browser UA → 302 to target_path
+ *   - Social crawler UA → 200 with OG <meta>, canonical = /s/<slug>
+ *   - Soft-deleted slug → 410 Gone (intentional-delete signal)
+ *   - Unknown slug → 404 Not Found
+ *   - Hit count increments after successful resolutions (both shapes)
+ *
+ * Mirrors the production server.js wiring but doesn't load the whole
+ * server — the surrounding middleware (CORS, helmet, rate limiters)
+ * isn't part of this route's contract.
+ */
+const express = require('express');
+const request = require('supertest');
+
+const { bootCrmDb } = require('./helpers/crmDb');
+
+jest.setTimeout(120000);
+
+let db; let cleanup; let service; let app;
+
+beforeAll(async () => {
+  ({ db, cleanup } = await bootCrmDb());
+
+  // Persist a business_profile + business_name so buildOgMetadata's
+  // settings-based fields populate consistently.
+  const { upsertAppSetting } = require('../../src/utils/appSettings');
+  await upsertAppSetting('branding_company_name', JSON.stringify('Test Studio'), 'string');
+
+  service = require('../../src/services/galleryShortUrlService');
+  const {
+    isSocialCrawler, buildOgMetadata, renderOgHtml,
+  } = require('../../src/services/galleryOgService');
+
+  app = express();
+  app.get('/s/:shortSlug', async (req, res) => {
+    try {
+      const row = await service.findByShortSlug(req.params.shortSlug);
+      if (!row) return res.status(404).type('text/plain').send('Short URL not found');
+      if (row.deleted_at) return res.status(410).type('text/plain').send('Short URL has been removed');
+
+      if (isSocialCrawler(req.get('user-agent'))) {
+        const event = await db('events').where({ id: row.event_id }).first('slug');
+        if (event?.slug) {
+          const meta = await buildOgMetadata(event.slug, req.originalUrl);
+          const base = (process.env.FRONTEND_URL || 'http://localhost:3000').replace(/\/$/, '');
+          meta.url = `${base}/s/${row.short_slug}`;
+          res.set('Cache-Control', 'public, max-age=300');
+          res.set('Content-Type', 'text/html; charset=utf-8');
+          res.send(renderOgHtml(meta));
+          service.recordHit(row.id).catch(() => {});
+          return;
+        }
+        return res.status(410).type('text/plain').send('Short URL points at a deleted event');
+      }
+
+      service.recordHit(row.id).catch(() => {});
+      return res.redirect(302, row.target_path);
+    } catch (err) {
+      return res.status(500).type('text/plain').send(err.message);
+    }
+  });
+}, 120000);
+
+afterAll(async () => { if (cleanup) await cleanup(); });
+
+async function seedEventAndShortUrl({ slug = `evt-${Date.now()}`, shortSlug }) {
+  const farFuture = new Date(Date.now() + 365 * 86400000).toISOString();
+  const [eventId] = await db('events').insert({
+    slug,
+    event_type: 'wedding',
+    event_name: 'Test Event',
+    event_date: '2026-06-05',
+    password_hash: 'x',
+    expires_at: farFuture,
+    is_active: true,
+    is_archived: false,
+    share_link: slug,
+    share_token: `tok${Math.random().toString(36).slice(2, 12)}`,
+    welcome_message: null,
+  });
+  const row = await service.createShortUrl({
+    eventId, customSlug: shortSlug,
+  });
+  return { eventId, shortUrl: row };
+}
+
+// User-agent strings the production `isSocialCrawler` helper matches.
+// Snapshot known-true samples here so the test stays in sync if the
+// helper's allowlist evolves.
+const BOT_UA_WHATSAPP = 'WhatsApp/2.23.20.0';
+const BOT_UA_FACEBOOK = 'facebookexternalhit/1.1 (+http://www.facebook.com/externalhit_uatext.php)';
+const BROWSER_UA = 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/605.1.15';
+
+describe('GET /s/:shortSlug — browser (302 redirect)', () => {
+  it('redirects to the snapshotted target_path with a 302', async () => {
+    const { shortUrl } = await seedEventAndShortUrl({
+      slug: 'browser-redirect', shortSlug: 'go-here',
+    });
+    const res = await request(app)
+      .get('/s/go-here')
+      .set('User-Agent', BROWSER_UA);
+    expect(res.status).toBe(302);
+    expect(res.headers.location).toBe(shortUrl.target_path);
+    expect(res.headers.location).toMatch(/^\/gallery\//);
+  });
+
+  it('increments hit_count on a browser hit (fire-and-forget — wait briefly)', async () => {
+    await seedEventAndShortUrl({
+      slug: 'hit-browser', shortSlug: 'hit-from-browser',
+    });
+    await request(app).get('/s/hit-from-browser').set('User-Agent', BROWSER_UA);
+    await new Promise((r) => setTimeout(r, 50));
+    const row = await service.findByShortSlug('hit-from-browser');
+    expect(row.hit_count).toBe(1);
+    expect(row.last_hit_at).toBeTruthy();
+  });
+});
+
+describe('GET /s/:shortSlug — social crawler (OG metadata)', () => {
+  it('returns 200 with OG HTML for WhatsApp UA', async () => {
+    await seedEventAndShortUrl({
+      slug: 'whatsapp-og', shortSlug: 'wa-preview',
+    });
+    const res = await request(app)
+      .get('/s/wa-preview')
+      .set('User-Agent', BOT_UA_WHATSAPP);
+    expect(res.status).toBe(200);
+    expect(res.headers['content-type']).toMatch(/text\/html/);
+    expect(res.text).toContain('<meta');
+    expect(res.text).toMatch(/og:title/);
+    expect(res.text).toMatch(/og:url/);
+  });
+
+  it('og:url canonical points at /s/<slug>, not the underlying gallery URL', async () => {
+    await seedEventAndShortUrl({
+      slug: 'canonical-test', shortSlug: 'canonical-short',
+    });
+    const res = await request(app)
+      .get('/s/canonical-short')
+      .set('User-Agent', BOT_UA_FACEBOOK);
+    expect(res.status).toBe(200);
+    // The og:url meta tag must contain the short-URL path, not the
+    // /gallery/<slug> path — this is the cache-key invariant from #699.
+    expect(res.text).toMatch(/property="og:url"\s+content="[^"]*\/s\/canonical-short"/);
+    expect(res.text).not.toMatch(
+      /property="og:url"\s+content="[^"]*\/gallery\/canonical-test"/
+    );
+  });
+
+  it('sets a short cache header so scrapers can re-fetch when admin rotates the preview', async () => {
+    await seedEventAndShortUrl({
+      slug: 'cache-header', shortSlug: 'cache-test',
+    });
+    const res = await request(app)
+      .get('/s/cache-test')
+      .set('User-Agent', BOT_UA_WHATSAPP);
+    expect(res.headers['cache-control']).toMatch(/public/);
+    expect(res.headers['cache-control']).toMatch(/max-age=300/);
+  });
+
+  it('increments hit_count on a crawler hit as well', async () => {
+    await seedEventAndShortUrl({
+      slug: 'hit-bot', shortSlug: 'hit-from-bot',
+    });
+    await request(app).get('/s/hit-from-bot').set('User-Agent', BOT_UA_WHATSAPP);
+    await new Promise((r) => setTimeout(r, 50));
+    const row = await service.findByShortSlug('hit-from-bot');
+    expect(row.hit_count).toBe(1);
+  });
+});
+
+describe('GET /s/:shortSlug — error states', () => {
+  it('404 for an unknown slug', async () => {
+    const res = await request(app)
+      .get('/s/never-existed')
+      .set('User-Agent', BROWSER_UA);
+    expect(res.status).toBe(404);
+  });
+
+  it('410 for a soft-deleted slug (intentional-delete signal)', async () => {
+    const { shortUrl } = await seedEventAndShortUrl({
+      slug: 'gone-test', shortSlug: 'gone-slug',
+    });
+    await service.softDelete(shortUrl.id, null);
+    const res = await request(app)
+      .get('/s/gone-slug')
+      .set('User-Agent', BROWSER_UA);
+    expect(res.status).toBe(410);
+  });
+
+  it('410 if the event was hard-deleted but the short URL row somehow survives', async () => {
+    const { eventId } = await seedEventAndShortUrl({
+      slug: 'orphan-test', shortSlug: 'orphan-slug',
+    });
+    // Hard-delete the event row (FK CASCADE would normally clean up the
+    // short URL too — but if CASCADE didn't fire for whatever reason
+    // (e.g. SQLite foreign_keys pragma off in a particular runtime), the
+    // resolver should still degrade safely).
+    // SQLite's foreign_keys pragma is OFF by default; the migration
+    // doesn't toggle it, so this delete leaves the short URL row.
+    await db('events').where({ id: eventId }).delete();
+    const res = await request(app)
+      .get('/s/orphan-slug')
+      .set('User-Agent', BOT_UA_WHATSAPP);
+    expect(res.status).toBe(410);
+  });
+
+  it('404 for a malformed slug (rejected at validation, no DB hit)', async () => {
+    const res = await request(app)
+      .get('/s/UPPER_CASE')
+      .set('User-Agent', BROWSER_UA);
+    expect(res.status).toBe(404);
+  });
+});
+
+describe('Regression — existing URL paths must still respond the same', () => {
+  // The /s/* namespace is additive: it must NOT shadow /gallery/*
+  // or any of the OG routes. We don't load the whole app here, but we
+  // can at least pin that the route param doesn't accept slashes —
+  // i.e. /s/foo/bar must NOT be matched by our handler.
+  it('the /s/:shortSlug route does not match nested paths', async () => {
+    const res = await request(app)
+      .get('/s/foo/bar')
+      .set('User-Agent', BROWSER_UA);
+    // Express returns its default 404 when no route matches the path.
+    expect(res.status).toBe(404);
+  });
+});

backend/__tests__/integration/galleryShortUrls.test.js

@@ -0,0 +1,282 @@
+/**
+ * Integration tests for the branded short-URL service (#699).
+ *
+ * Exercises createShortUrl + findByShortSlug + listForEvent + softDelete
+ * + recordHit against a real SQLite DB, including the contracts that
+ * matter for production correctness:
+ *
+ *   - Custom slug + collision detection (409 with `suggested`)
+ *   - Auto-generated slug from event slug + year
+ *   - Soft-delete preserves the row (admin can audit)
+ *   - target_path snapshots at create time (toggling the global
+ *     "Use short gallery URLs" setting later doesn't change existing
+ *     short URLs — backward-compat invariant from #699)
+ *   - hit_count increments idempotently
+ *   - findByShortSlug returns soft-deleted rows (caller decides 410 vs 404)
+ *
+ * Boots one DB for the whole file (cheap on SQLite); each test seeds
+ * its own event row to keep scope clean.
+ */
+const { bootCrmDb } = require('./helpers/crmDb');
+
+jest.setTimeout(120000);
+
+let db; let cleanup; let service; let adminId;
+
+beforeAll(async () => {
+  ({ db, cleanup } = await bootCrmDb());
+
+  // Minimal admin for created_by audit.
+  const adminInsert = await db('admin_users').insert({
+    username: 'shorturl-test',
+    email: 'shorturl@example.com',
+    password_hash: 'x',
+    must_change_password: false,
+    created_at: new Date(),
+  }).returning('id');
+  adminId = adminInsert[0]?.id ?? adminInsert[0];
+
+  service = require('../../src/services/galleryShortUrlService');
+}, 120000);
+
+afterAll(async () => { if (cleanup) await cleanup(); });
+
+// Each test seeds a fresh event so collisions / counter state don't leak.
+async function seedEvent(overrides = {}) {
+  const slug = overrides.slug || `evt-${Date.now()}-${Math.random().toString(36).slice(2, 6)}`;
+  const farFuture = new Date(Date.now() + 365 * 86400000).toISOString();
+  const [id] = await db('events').insert({
+    slug,
+    event_type: 'wedding',
+    event_name: overrides.event_name || 'Test Wedding',
+    event_date: overrides.event_date || '2026-06-05',
+    password_hash: 'x',
+    expires_at: farFuture,
+    is_active: true,
+    is_archived: false,
+    share_link: slug,
+    share_token: overrides.share_token || `tok${Math.random().toString(36).slice(2, 12)}`,
+    welcome_message: null,
+  });
+  const event = await db('events').where({ id }).first();
+  return event;
+}
+
+describe('createShortUrl — custom slug', () => {
+  it('creates with a custom slug', async () => {
+    const event = await seedEvent({ slug: 'sofia-grad-1' });
+    const row = await service.createShortUrl({
+      eventId: event.id,
+      customSlug: 'sofia-graduation-1',
+      createdBy: adminId,
+    });
+    expect(row.short_slug).toBe('sofia-graduation-1');
+    expect(row.target_path).toBe(`/gallery/${event.slug}`);
+    expect(row.event_id).toBe(event.id);
+    expect(row.hit_count).toBe(0);
+  });
+
+  it('lowercases the input — operators pasting mixed-case still get a clean slug', async () => {
+    const event = await seedEvent({ slug: 'sofia-grad-2' });
+    const row = await service.createShortUrl({
+      eventId: event.id,
+      customSlug: 'Sofia-GraduAtion-2',  // mixed case
+      createdBy: adminId,
+    });
+    expect(row.short_slug).toBe('sofia-graduation-2');
+  });
+
+  it('rejects an invalid slug with INVALID_SLUG code', async () => {
+    const event = await seedEvent({ slug: 'invalid-test' });
+    await expect(service.createShortUrl({
+      eventId: event.id,
+      customSlug: 'invalid slug with spaces',
+      createdBy: adminId,
+    })).rejects.toMatchObject({ code: 'INVALID_SLUG' });
+  });
+
+  it('rejects a reserved slug with INVALID_SLUG code', async () => {
+    const event = await seedEvent({ slug: 'reserved-test' });
+    await expect(service.createShortUrl({
+      eventId: event.id,
+      customSlug: 'admin',
+      createdBy: adminId,
+    })).rejects.toMatchObject({ code: 'INVALID_SLUG' });
+  });
+
+  it('rejects a duplicate slug with SLUG_TAKEN + suggested fallback', async () => {
+    const event1 = await seedEvent({ slug: 'dup-test-1' });
+    const event2 = await seedEvent({ slug: 'dup-test-2' });
+    await service.createShortUrl({ eventId: event1.id, customSlug: 'collide-me' });
+    await expect(service.createShortUrl({
+      eventId: event2.id, customSlug: 'collide-me',
+    })).rejects.toMatchObject({
+      code: 'SLUG_TAKEN',
+      suggested: expect.any(String),
+    });
+  });
+
+  it('throws EVENT_NOT_FOUND when the event id does not exist', async () => {
+    await expect(service.createShortUrl({
+      eventId: 9999999, customSlug: 'no-event',
+    })).rejects.toMatchObject({ code: 'EVENT_NOT_FOUND' });
+  });
+});
+
+describe('createShortUrl — auto-generated slug', () => {
+  it('uses event slug + year when no custom slug provided', async () => {
+    const event = await seedEvent({
+      slug: 'autogen-wedding', event_date: '2026-06-05',
+    });
+    const row = await service.createShortUrl({
+      eventId: event.id,
+      createdBy: adminId,
+    });
+    // First-choice candidate is just the slug; takes that.
+    expect(row.short_slug).toBe('autogen-wedding');
+  });
+
+  it('falls back to slug-year when the bare slug is already taken', async () => {
+    // Both events SHARE the same canonical slug so the first-choice
+    // bare-slug candidate is burned, forcing autoGen to try the
+    // year-suffixed variant.
+    const event1 = await seedEvent({
+      slug: 'collide-base', event_date: '2026-07-01',
+    });
+    await service.createShortUrl({
+      eventId: event1.id, customSlug: 'collide-base',
+    });
+    const event2 = await seedEvent({
+      slug: 'collide-base-2', event_date: '2026-07-01',
+    });
+    // Force the bare candidate of event2 to also collide by burning it.
+    await service.createShortUrl({
+      eventId: event1.id, customSlug: 'collide-base-2',
+    });
+    const row = await service.createShortUrl({
+      eventId: event2.id,  // No custom — auto-gen from event2.slug
+    });
+    // Bare candidate `collide-base-2` is taken → year-suffixed picks.
+    expect(row.short_slug).toBe('collide-base-2-2026');
+  });
+});
+
+describe('createShortUrl — target_path snapshotting (#699 backward-compat)', () => {
+  it('uses /gallery/<slug> when the global short-URLs setting is OFF (default)', async () => {
+    const event = await seedEvent({ slug: 'snapshot-off' });
+    const row = await service.createShortUrl({
+      eventId: event.id, customSlug: 'snap-off',
+    });
+    expect(row.target_path).toBe(`/gallery/${event.slug}`);
+  });
+
+  it('uses /gallery/<share_token> when the global setting is ON at create time', async () => {
+    // Persist the setting.
+    const { upsertAppSetting } = require('../../src/utils/appSettings');
+    await upsertAppSetting('general_use_short_gallery_urls', JSON.stringify(true), 'system');
+    try {
+      const event = await seedEvent({ slug: 'snapshot-on', share_token: 'tokenAbc123' });
+      const row = await service.createShortUrl({
+        eventId: event.id, customSlug: 'snap-on',
+      });
+      expect(row.target_path).toBe(`/gallery/${event.share_token}`);
+
+      // CRITICAL backward-compat invariant: now flip the setting OFF.
+      // Existing short URLs must still resolve to the same target_path
+      // they were created with — operator's existing share links don't
+      // silently change behaviour.
+      await upsertAppSetting('general_use_short_gallery_urls', JSON.stringify(false), 'system');
+      const refetched = await service.findByShortSlug('snap-on');
+      expect(refetched.target_path).toBe(`/gallery/${event.share_token}`);
+    } finally {
+      await upsertAppSetting('general_use_short_gallery_urls', JSON.stringify(false), 'system');
+    }
+  });
+});
+
+describe('findByShortSlug + listForEvent', () => {
+  it('returns null for an unknown slug', async () => {
+    expect(await service.findByShortSlug('does-not-exist-xyz')).toBeNull();
+  });
+
+  it('returns null for a malformed slug (no DB hit)', async () => {
+    expect(await service.findByShortSlug('UPPER_CASE')).toBeNull();
+    expect(await service.findByShortSlug('with spaces')).toBeNull();
+    expect(await service.findByShortSlug('')).toBeNull();
+  });
+
+  it('returns soft-deleted rows (caller decides 410 vs 404)', async () => {
+    const event = await seedEvent({ slug: 'softdel-find' });
+    const created = await service.createShortUrl({
+      eventId: event.id, customSlug: 'find-deleted',
+    });
+    await service.softDelete(created.id, adminId);
+    const fetched = await service.findByShortSlug('find-deleted');
+    expect(fetched).not.toBeNull();
+    expect(fetched.deleted_at).toBeTruthy();
+  });
+
+  it('listForEvent excludes soft-deleted rows', async () => {
+    const event = await seedEvent({ slug: 'list-test' });
+    const live = await service.createShortUrl({
+      eventId: event.id, customSlug: 'list-live',
+    });
+    const deleted = await service.createShortUrl({
+      eventId: event.id, customSlug: 'list-deleted',
+    });
+    await service.softDelete(deleted.id, adminId);
+    const list = await service.listForEvent(event.id);
+    const ids = list.map((r) => r.id);
+    expect(ids).toContain(live.id);
+    expect(ids).not.toContain(deleted.id);
+  });
+});
+
+describe('softDelete', () => {
+  it('returns true on first call, false on second (idempotent admin clicks)', async () => {
+    const event = await seedEvent({ slug: 'softdel-idem' });
+    const created = await service.createShortUrl({
+      eventId: event.id, customSlug: 'idem-delete',
+    });
+    expect(await service.softDelete(created.id, adminId)).toBe(true);
+    expect(await service.softDelete(created.id, adminId)).toBe(false);
+  });
+
+  it('returns false for an unknown id (caller maps to 404)', async () => {
+    expect(await service.softDelete(9999999, adminId)).toBe(false);
+  });
+});
+
+describe('createShortUrl after soft-delete — slug rotation', () => {
+  it('re-creating a soft-deleted slug succeeds (purges the deleted row)', async () => {
+    const event = await seedEvent({ slug: 'rotate' });
+    const first = await service.createShortUrl({
+      eventId: event.id, customSlug: 'rotate-me',
+    });
+    await service.softDelete(first.id, adminId);
+    // The slug is now reclaimable for a fresh row.
+    const second = await service.createShortUrl({
+      eventId: event.id, customSlug: 'rotate-me',
+    });
+    expect(second.id).not.toBe(first.id);
+    expect(second.short_slug).toBe('rotate-me');
+  });
+});
+
+describe('recordHit', () => {
+  it('increments hit_count + stamps last_hit_at', async () => {
+    const event = await seedEvent({ slug: 'hit-counter' });
+    const row = await service.createShortUrl({
+      eventId: event.id, customSlug: 'count-me',
+    });
+    await service.recordHit(row.id);
+    await service.recordHit(row.id);
+    const fetched = await service.findByShortSlug('count-me');
+    expect(fetched.hit_count).toBe(2);
+    expect(fetched.last_hit_at).toBeTruthy();
+  });
+
+  it('is fire-and-forget — invalid id does not throw', async () => {
+    await expect(service.recordHit(9999999)).resolves.not.toThrow();
+  });
+});

backend/__tests__/integration/helpers/crmDb.js

@@ -0,0 +1,230 @@
+/**
+ * Test harness for CRM integration tests.
+ *
+ * Boots a temp-SQLite database, runs every `migrations/core/*.up()`
+ * directly (bypassing knex's Migrator — its exclusive write lock
+ * deadlocks 001_init's nested `initializeDatabase()` call), and
+ * exposes a small helper for seeding the minimal row set that the
+ * quote/contract/invoice services need to operate.
+ *
+ * Usage:
+ *
+ *   const { bootCrmDb, seedMinimal } = require('./helpers/crmDb');
+ *
+ *   beforeAll(async () => {
+ *     ({ db, cleanup } = await bootCrmDb());
+ *     ({ adminId, customerId } = await seedMinimal(db));
+ *   });
+ *   afterAll(async () => { await cleanup(); });
+ */
+
+const fs = require('fs');
+const path = require('path');
+const os = require('os');
+const bcrypt = require('bcrypt');
+
+async function runCoreMigrations(db) {
+  await db.schema.createTable('migrations', (t) => {
+    t.increments('id').primary();
+    t.string('filename').unique().notNullable();
+    t.timestamp('applied_at').defaultTo(db.fn.now());
+  });
+
+  const coreDir = path.resolve(__dirname, '..', '..', '..', 'migrations', 'core');
+  const files = (await fs.promises.readdir(coreDir))
+    .filter((f) => f.endsWith('.js'))
+    .sort();
+
+  for (const f of files) {
+    const mod = require(path.join(coreDir, f));
+    if (typeof mod.up === 'function') {
+      await mod.up(db);
+    }
+    await db('migrations').insert({ filename: f });
+  }
+}
+
+/**
+ * Boot a clean test DB. Returns { db, cleanup, tmpDir }.
+ * Caller must invoke cleanup() in afterAll to release the SQLite file
+ * and the temp directory.
+ */
+async function bootCrmDb() {
+  const tmpDir = await fs.promises.mkdtemp(path.join(os.tmpdir(), 'picpeak-crm-'));
+  process.env.NODE_ENV = 'test';
+  process.env.TEST_DATABASE_PATH = path.join(tmpDir, 'crm.db');
+  process.env.STORAGE_PATH = path.join(tmpDir, 'storage');
+  await fs.promises.mkdir(process.env.STORAGE_PATH, { recursive: true });
+
+  // No jest.resetModules() — every service the test later requires
+  // must share THIS db instance. Two module copies on one SQLite file
+  // each open their own knex pool and the SQLite write lock deadlocks
+  // the second one acquiring a connection. Caller is responsible for
+  // setting TEST_DATABASE_PATH before the first require of db.js
+  // (which knexfile reads at module-init time); bootCrmDb only works
+  // when invoked before any service import.
+  const { db } = require('../../../src/database/db');
+
+  await runCoreMigrations(db);
+
+  return {
+    db,
+    tmpDir,
+    cleanup: async () => {
+      try { await db.destroy(); } catch (_) {}
+      try { await fs.promises.rm(tmpDir, { recursive: true, force: true }); } catch (_) {}
+    },
+  };
+}
+
+/**
+ * Seed the minimal row set that quote/contract/invoice services
+ * dereference on creation: an admin user, an active customer, a
+ * business_profile row, and the app_settings keys the services read.
+ *
+ * Returns the ids the caller will pass into service calls.
+ */
+async function seedMinimal(db) {
+  const passwordHash = await bcrypt.hash('test-pass', 4); // low rounds = fast
+
+  const adminInsert = await db('admin_users').insert({
+    username: 'tester', email: 'tester@example.com',
+    password_hash: passwordHash, must_change_password: false,
+    created_at: new Date(),
+  }).returning('id');
+  const adminId = adminInsert[0]?.id ?? adminInsert[0];
+
+  // business_profile is a singleton; the row is seeded by migration 107
+  // for fresh installs. Defensive: insert if missing.
+  const profile = await db('business_profile').first();
+  if (!profile) {
+    await db('business_profile').insert({
+      legal_name: 'Test Studio',
+      default_currency: 'CHF',
+      default_locale: 'de',
+    });
+  }
+
+  const customerInsert = await db('customer_accounts').insert({
+    email: 'customer@example.com',
+    display_name: 'Test Customer',
+    password_hash: passwordHash,
+    preferred_language: 'de',
+    is_active: 1,
+    created_at: new Date(),
+  }).returning('id');
+  const customerId = customerInsert[0]?.id ?? customerInsert[0];
+
+  return { adminId, customerId };
+}
+
+// ---------------------------------------------------------------------
+// Route-test helpers (#570) — building blocks for the CRM HTTP layer
+// tests. Kept here so every supertest suite shares the same minting +
+// app-wiring shape and a refactor lands in one place.
+// ---------------------------------------------------------------------
+
+const crypto = require('crypto');
+const jwt = require('jsonwebtoken');
+const express = require('express');
+const cookieParser = require('cookie-parser');
+
+/**
+ * Promote a seeded admin into a role (default `super_admin`) so
+ * `requirePermission(...)` checks pass. seedMinimal creates an admin
+ * without a role — that's good for negative tests (expect 403) but
+ * happy-path tests need the role assignment.
+ *
+ * Returns the role id the admin was assigned to.
+ */
+async function assignAdminRole(db, adminId, roleName = 'super_admin') {
+  const role = await db('roles').where({ name: roleName }).first();
+  if (!role) {
+    throw new Error(`Role '${roleName}' not seeded — check the test DB`);
+  }
+  await db('admin_users').where({ id: adminId }).update({ role_id: role.id });
+  return role.id;
+}
+
+/**
+ * Mint an admin JWT in the same shape adminAuth middleware expects.
+ * The tests inject this via `Authorization: Bearer <token>`.
+ */
+function mintAdminToken(adminId, { expiresIn = '1h', extraClaims = {} } = {}) {
+  process.env.JWT_SECRET = process.env.JWT_SECRET || 'crm-route-test-secret';
+  return jwt.sign(
+    { id: adminId, type: 'admin', iat: Math.floor(Date.now() / 1000), ...extraClaims },
+    process.env.JWT_SECRET,
+    { expiresIn, issuer: 'picpeak-auth' }
+  );
+}
+
+/**
+ * Insert a row into one of the public-token tables for testing the
+ * loadActionToken guard outcomes. Returns the generated 64-hex token.
+ *
+ * Usage:
+ *   await createPublicToken(db, 'quote_action_tokens', { quote_id: q.id });
+ *   await createPublicToken(db, 'quote_action_tokens', { quote_id: q.id, expires_at: pastDate });
+ *   await createPublicToken(db, 'quote_action_tokens', { quote_id: q.id, used_at: new Date() });
+ *   await createPublicToken(db, 'quote_action_tokens', { quote_id: q.id, expires_at: null });
+ */
+async function createPublicToken(db, tableName, opts = {}) {
+  const token = opts.token || crypto.randomBytes(32).toString('hex');
+  const expiresAt = opts.expires_at === null
+    ? null
+    : (opts.expires_at || new Date(Date.now() + 7 * 24 * 60 * 60 * 1000));
+  // Serialise Date → ISO string. Bare Date objects round-tripped
+  // inconsistently through knex+SQLite — sometimes as epoch ms,
+  // sometimes via .toString() → literal "[object Object]" which then
+  // parses back to NaN and silently defeats the expiry guard.
+  const toStorable = (v) => (v instanceof Date ? v.toISOString() : v);
+  const row = {
+    ...opts,
+    token,
+    expires_at: toStorable(expiresAt),
+    created_at: toStorable(new Date()),
+  };
+  await db(tableName).insert(row);
+  return token;
+}
+
+/**
+ * Build an Express app with the requested route file mounted. Mirrors
+ * the production app's middleware shape (json + cookies) but skips
+ * everything else (CORS, helmet, rate limiters) — route tests pin the
+ * handler's contract, not the surrounding cross-cutting concerns.
+ *
+ * Example:
+ *   const app = buildRouteApp('/api/public/quotes',
+ *     require('../../src/routes/publicQuotes'));
+ */
+function buildRouteApp(mount, router) {
+  const app = express();
+  app.use(express.json());
+  app.use(cookieParser());
+  app.use(mount, router);
+  // Catch-all error handler. Mirrors the real middleware/errorHandler:
+  // AppError subclasses (ValidationError, NotFoundError, etc.) use
+  // `.statusCode` (NOT `.status` — getting that wrong silently maps
+  // every 400 / 404 / 410 to 500 in tests).
+  // eslint-disable-next-line no-unused-vars
+  app.use((err, req, res, next) => {
+    const statusCode = err.statusCode || err.status || 500;
+    res.status(statusCode).json({
+      error: err.message || 'Internal error',
+      code: err.code,
+      ...(err.details ? { details: err.details } : {}),
+    });
+  });
+  return app;
+}
+
+module.exports = {
+  bootCrmDb,
+  seedMinimal,
+  assignAdminRole,
+  mintAdminToken,
+  createPublicToken,
+  buildRouteApp,
+};

backend/__tests__/integration/hiddenOwnFeedback.test.js

@@ -0,0 +1,245 @@
+/**
+ * Hidden feedback, seen from the guest who left it (#1150).
+ *
+ * Everything in the system treats a hidden row as absent: getPhotoFeedback
+ * drops it even for the guest's own feedback, the /photos filters drop it, and
+ * updatePhotoFeedbackStats does not count it. One place disagreed — the
+ * per-viewer `is_liked` heart — so a like the photographer had hidden still
+ * showed as liked on a photo whose like_count was zero. (The `my_color_label`
+ * badge has the same shape on main; colour labels are not on this branch.)
+ *
+ * Making those two agree exposes the second half: the duplicate check that
+ * powers like/favorite toggling did NOT skip hidden rows, so the now-empty
+ * heart, when clicked, found the hidden row and toggled it OFF. The click
+ * appeared to do nothing and it took two more to get back to a filled heart.
+ *
+ * Hiding a non-comment is deliberate, not an accident of the raw route: #839
+ * and #1044 both ship it, with tests asserting that a hidden reaction or
+ * colour label stops counting. So the fix is to make hidden mean absent
+ * consistently — not to stop admins hiding these.
+ */
+
+const request = require('supertest');
+const express = require('express');
+const cookieParser = require('cookie-parser');
+const jwt = require('jsonwebtoken');
+
+const { bootCrmDb, seedMinimal } = require('./helpers/crmDb');
+
+process.env.JWT_SECRET = process.env.JWT_SECRET || 'hidden-feedback-secret';
+
+const SLUG = 'hidden-own-feedback';
+const ME = 'guest-me-identifier';
+
+describe('a guest\'s own hidden feedback (#1150)', () => {
+  let db; let cleanup; let app; let feedbackService;
+  let eventId; let photoId; let myGuestRowId;
+
+  const galleryToken = () => jwt.sign(
+    { eventId, eventSlug: SLUG, type: 'gallery' },
+    process.env.JWT_SECRET,
+    { expiresIn: '1h', issuer: 'picpeak-auth' }
+  );
+  const guestToken = () => jwt.sign(
+    { type: 'guest', guestId: myGuestRowId, eventId },
+    process.env.JWT_SECRET,
+    { expiresIn: '1h', issuer: 'picpeak-auth' }
+  );
+
+  const getPhoto = async () => {
+    const res = await request(app)
+      .get(`/api/gallery/${SLUG}/photos`)
+      .set('Authorization', `Bearer ${galleryToken()}`)
+      .set('x-guest-token', guestToken());
+    expect(res.status).toBe(200);
+    const photos = Array.isArray(res.body) ? res.body : res.body.photos;
+    return (photos || []).find((p) => p.id === photoId);
+  };
+
+  beforeAll(async () => {
+    ({ db, cleanup } = await bootCrmDb());
+    await seedMinimal(db);
+    feedbackService = require('../../src/services/feedbackService');
+
+    const [ev] = await db('events').insert({
+      slug: SLUG,
+      event_type: 'wedding',
+      event_name: 'Hidden Own Feedback',
+      event_date: '2026-08-01',
+      host_email: 'h@example.com',
+      admin_email: 'a@example.com',
+      password_hash: 'x',
+      share_link: `/gallery/${SLUG}/share`,
+      share_token: 'hidden-own-share',
+      expires_at: new Date(Date.now() + 7 * 24 * 3600 * 1000).toISOString(),
+      is_active: 1, is_archived: 0, is_draft: 0,
+      created_at: new Date().toISOString(),
+    }).returning('id');
+    eventId = typeof ev === 'object' ? ev.id : ev;
+
+    const [p] = await db('photos').insert({
+      event_id: eventId, filename: 'shot.jpg', path: 'events/hidden/shot.jpg',
+      type: 'individual', uploaded_at: new Date().toISOString(),
+    }).returning('id');
+    photoId = typeof p === 'object' ? p.id : p;
+
+    const [g] = await db('gallery_guests').insert({
+      event_id: eventId, name: 'Me', identifier: ME,
+      created_at: new Date().toISOString(), last_seen_at: new Date().toISOString(),
+      is_deleted: false,
+    }).returning('id');
+    myGuestRowId = typeof g === 'object' ? g.id : g;
+
+    await db('event_feedback_settings').insert({
+      event_id: eventId, feedback_enabled: true, allow_likes: true,
+      moderate_comments: false,
+      show_feedback_to_guests: true,
+    });
+
+    app = express();
+    app.use(express.json());
+    app.use(cookieParser());
+    app.use('/api/gallery', require('../../src/routes/gallery'));
+    app.use('/api/gallery', require('../../src/routes/galleryFeedback'));
+  }, 180000);
+
+  afterAll(async () => { if (cleanup) await cleanup(); });
+
+  const like = () => db('photo_feedback').insert({
+    photo_id: photoId, event_id: eventId, guest_identifier: ME,
+    guest_id: myGuestRowId, feedback_type: 'like',
+    is_approved: true, is_hidden: false, created_at: new Date().toISOString(),
+  });
+
+  beforeEach(async () => {
+    await db('photo_feedback').where({ photo_id: photoId }).del();
+    await db('photos').where('id', photoId).update({ like_count: 0 });
+  });
+
+  describe('the read surfaces agree with each other', () => {
+    it('un-fills the heart once the like is hidden', async () => {
+      await like();
+      await feedbackService.updatePhotoFeedbackStats(photoId);
+      expect((await getPhoto()).is_liked).toBe(true);
+
+      await db('photo_feedback')
+        .where({ photo_id: photoId, feedback_type: 'like' })
+        .update({ is_hidden: true });
+      await feedbackService.updatePhotoFeedbackStats(photoId);
+
+      const photo = await getPhoto();
+      // like_count already ignored hidden rows, so the heart was the only
+      // thing still claiming this photo was liked.
+      expect(photo.like_count).toBe(0);
+      expect(photo.is_liked).toBe(false);
+    });
+
+  });
+
+  describe('and every other surface agrees', () => {
+    it('keeps a hidden like out of /my-feedback', async () => {
+      await like();
+      await db('photo_feedback')
+        .where({ photo_id: photoId, feedback_type: 'like' })
+        .update({ is_hidden: true });
+
+      const res = await request(app)
+        .get(`/api/gallery/${SLUG}/my-feedback`)
+        .set('Authorization', `Bearer ${galleryToken()}`)
+        .set('x-guest-token', guestToken());
+      expect(res.status).toBe(200);
+
+      // In guest identity mode the Liked/Favorited/Rated chips and their
+      // filters are built from THIS array, not from is_liked — so a hidden
+      // like left an empty heart while the chip still counted it.
+      expect(res.body.filter((f) => f.feedback_type === 'like')).toHaveLength(0);
+    });
+
+    it('does not count a hidden row against the guest cap', async () => {
+      await db('event_feedback_settings')
+        .where({ event_id: eventId }).update({ max_likes_per_guest: 1 });
+      await like();
+      await db('photo_feedback')
+        .where({ photo_id: photoId, feedback_type: 'like' })
+        .update({ is_hidden: true });
+
+      // The hidden row is room, not an occupant: the guest sees an empty
+      // heart, and meeting that click with limit_reached leaves the control
+      // dead until they un-like something they can still see.
+      const result = await feedbackService.submitFeedback(photoId, eventId, {
+        feedback_type: 'like', guest_identifier: ME, guest_id: myGuestRowId,
+      });
+      expect(result.limit_reached).toBeUndefined();
+
+      await db('event_feedback_settings')
+        .where({ event_id: eventId }).update({ max_likes_per_guest: null });
+    });
+
+    it('leaves other anonymous rows alone when there is no identity to scope by', async () => {
+      // With neither guest_id nor guest_identifier the collapse scope degrades
+      // to `guest_identifier IS NULL` — every identifier-less row on the
+      // photo, i.e. other people's.
+      const anon = (extra) => ({
+        photo_id: photoId, event_id: eventId, feedback_type: 'like',
+        is_approved: true, created_at: new Date().toISOString(), ...extra,
+      });
+      const [h] = await db('photo_feedback').insert(anon({ is_hidden: true })).returning('id');
+      const hiddenId = typeof h === 'object' ? h.id : h;
+      await db('photo_feedback').insert(anon({ is_hidden: false }));
+      await db('photo_feedback').insert(anon({ is_hidden: false }));
+
+      await feedbackService.moderateFeedback(hiddenId, 'approve', 1);
+
+      expect(await db('photo_feedback')
+        .where({ photo_id: photoId, feedback_type: 'like', is_hidden: false }))
+        .toHaveLength(3);
+    });
+
+    it('collapses the replacement when an admin unhides the original', async () => {
+      await like();
+      const original = await db('photo_feedback').where({ photo_id: photoId }).first();
+      await db('photo_feedback').where('id', original.id).update({ is_hidden: true });
+
+      await feedbackService.submitFeedback(photoId, eventId, {
+        feedback_type: 'like', guest_identifier: ME, guest_id: myGuestRowId,
+      });
+      expect(await db('photo_feedback').where({ photo_id: photoId })).toHaveLength(2);
+
+      await feedbackService.moderateFeedback(original.id, 'approve', 1);
+
+      // Two visible rows for one guest would double-count in the tallies and
+      // need two toggles to clear, since each deletes a single row.
+      const visible = await db('photo_feedback')
+        .where({ photo_id: photoId, feedback_type: 'like', is_hidden: false });
+      expect(visible).toHaveLength(1);
+      expect(visible[0].id).toBe(original.id);
+    });
+  });
+
+  describe('and clicking still works afterwards', () => {
+    it('re-liking creates a fresh row instead of toggling the hidden one off', async () => {
+      await like();
+      await db('photo_feedback')
+        .where({ photo_id: photoId, feedback_type: 'like' })
+        .update({ is_hidden: true });
+
+      // What the guest sees is an empty heart, so this is an ADD.
+      const result = await feedbackService.submitFeedback(photoId, eventId, {
+        feedback_type: 'like',
+        guest_identifier: ME,
+        guest_id: myGuestRowId,
+      });
+
+      // Before this, the duplicate check found the hidden row and deleted it —
+      // `removed: true` — so the click did nothing visible and the moderation
+      // was silently undone.
+      expect(result.removed).toBeUndefined();
+
+      const visible = await db('photo_feedback')
+        .where({ photo_id: photoId, feedback_type: 'like', is_hidden: false });
+      expect(visible).toHaveLength(1);
+      expect((await getPhoto()).is_liked).toBe(true);
+    });
+  });
+
+});

backend/__tests__/integration/imageProcessor.storage.test.js

@@ -0,0 +1,200 @@
+const path = require('path');
+const fs = require('fs').promises;
+const fsSync = require('fs');
+const os = require('os');
+const crypto = require('crypto');
+const { S3Client, CreateBucketCommand, DeleteBucketCommand, ListObjectsV2Command, DeleteObjectsCommand } = require('@aws-sdk/client-s3');
+const sharp = require('sharp');
+
+const LocalFsStorage = require('../../src/services/storage/LocalFsStorage');
+const S3StorageBackend = require('../../src/services/storage/S3StorageBackend');
+const storageModule = require('../../src/services/storage');
+
+// Stub out the DB so getThumbnailSettings falls into its catch and uses defaults.
+jest.mock('../../src/database/db', () => ({
+  db: () => {
+    throw new Error('db disabled in this test');
+  },
+}));
+
+const TEST_S3 = {
+  endpoint: process.env.TEST_S3_ENDPOINT || 'http://localhost:7104',
+  accessKeyId: process.env.TEST_S3_ACCESS_KEY || 'minioadmin',
+  secretAccessKey: process.env.TEST_S3_SECRET_KEY || 'minioadmin',
+  region: 'us-east-1',
+};
+
+const skipS3 = process.env.SKIP_S3_TESTS === 'true';
+
+function backendCases() {
+  const cases = [
+    {
+      name: 'LocalFsStorage',
+      async setup() {
+        const root = await fs.mkdtemp(path.join(os.tmpdir(), 'picpeak-imgproc-'));
+        const storage = new LocalFsStorage({ root });
+        await storage.init();
+        return { storage, cleanup: () => fs.rm(root, { recursive: true, force: true }) };
+      },
+    },
+  ];
+  if (!skipS3) {
+    cases.push({
+      name: 'S3StorageBackend (MinIO)',
+      async setup() {
+        const bucket = `picpeak-imgproc-${Date.now()}-${crypto.randomBytes(2).toString('hex')}`;
+        const s3Client = new S3Client({
+          endpoint: TEST_S3.endpoint,
+          region: TEST_S3.region,
+          credentials: { accessKeyId: TEST_S3.accessKeyId, secretAccessKey: TEST_S3.secretAccessKey },
+          forcePathStyle: true,
+        });
+        await s3Client.send(new CreateBucketCommand({ Bucket: bucket }));
+        const storage = new S3StorageBackend({
+          bucket,
+          region: TEST_S3.region,
+          endpoint: TEST_S3.endpoint,
+          accessKeyId: TEST_S3.accessKeyId,
+          secretAccessKey: TEST_S3.secretAccessKey,
+          forcePathStyle: true,
+          sslEnabled: false,
+        });
+        await storage.init();
+        return {
+          storage,
+          async cleanup() {
+            const list = await s3Client.send(new ListObjectsV2Command({ Bucket: bucket }));
+            if (list.Contents?.length) {
+              await s3Client.send(new DeleteObjectsCommand({
+                Bucket: bucket,
+                Delete: { Objects: list.Contents.map((o) => ({ Key: o.Key })) },
+              }));
+            }
+            await s3Client.send(new DeleteBucketCommand({ Bucket: bucket }));
+          },
+        };
+      },
+    });
+  }
+  return cases;
+}
+
+async function makeSourceJpeg(targetDir, name) {
+  const localPath = path.join(targetDir, name);
+  // 800x600 random RGB image so sharp has something realistic to thumbnail.
+  const width = 800;
+  const height = 600;
+  const buf = Buffer.alloc(width * height * 3);
+  for (let i = 0; i < buf.length; i++) buf[i] = (i * 7) % 256;
+  await sharp(buf, { raw: { width, height, channels: 3 } })
+    .jpeg({ quality: 90 })
+    .toFile(localPath);
+  return localPath;
+}
+
+describe.each(backendCases())('imageProcessor through $name', ({ setup }) => {
+  let storage;
+  let cleanup;
+  let tmpDir;
+  let imageProcessor;
+
+  beforeAll(async () => {
+    ({ storage, cleanup } = await setup());
+    storageModule.setStorageForTesting(storage);
+    // Require AFTER setStorageForTesting so the module sees our injection.
+    delete require.cache[require.resolve('../../src/services/imageProcessor')];
+    imageProcessor = require('../../src/services/imageProcessor');
+    tmpDir = await fs.mkdtemp(path.join(os.tmpdir(), 'picpeak-imgproc-src-'));
+  }, 30000);
+
+  afterAll(async () => {
+    storageModule.resetStorage();
+    if (tmpDir) await fs.rm(tmpDir, { recursive: true, force: true }).catch(() => {});
+    if (cleanup) await cleanup();
+  });
+
+  test('generateThumbnail writes through storage and returns a relative key', async () => {
+    const src = await makeSourceJpeg(tmpDir, 'sample.jpg');
+    const key = await imageProcessor.generateThumbnail(src);
+    expect(key).toBe('thumbnails/thumb_sample.jpg');
+
+    expect(await storage.exists(key)).toBe(true);
+    const stat = await storage.stat(key);
+    expect(stat.size).toBeGreaterThan(100);
+
+    // Verify the bytes are a valid JPEG by re-parsing with sharp on local mode.
+    if (storage.kind() === 'local') {
+      const meta = await sharp(storage.resolveLocalPath(key)).metadata();
+      expect(meta.format).toBe('jpeg');
+      expect(meta.width).toBeLessThanOrEqual(300);
+    }
+  });
+
+  test('generateHeroImage writes through storage and returns a relative key', async () => {
+    const src = await makeSourceJpeg(tmpDir, 'hero-source.jpg');
+    const key = await imageProcessor.generateHeroImage(src);
+    expect(key).toBe('heroes/hero_hero-source.jpg');
+    expect(await storage.exists(key)).toBe(true);
+  });
+
+  test('generatePreviewImage writes to /previews and skips enlargement of small originals', async () => {
+    const src = await makeSourceJpeg(tmpDir, 'preview-source.jpg');
+    const key = await imageProcessor.generatePreviewImage(src);
+    expect(key).toBe('previews/preview_preview-source.jpg');
+    expect(await storage.exists(key)).toBe(true);
+
+    if (storage.kind() === 'local') {
+      const meta = await sharp(storage.resolveLocalPath(key)).metadata();
+      expect(meta.format).toBe('jpeg');
+      // Source is 800x600 and default longEdge is 1920 with
+      // withoutEnlargement: true → preview must NOT be upscaled.
+      expect(meta.width).toBe(800);
+      expect(meta.height).toBe(600);
+    }
+  });
+
+  test('generatePreviewImage shrinks oversized images to fit longEdge while preserving aspect', async () => {
+    const src = await makeSourceJpeg(tmpDir, 'preview-shrink.jpg');
+    const key = await imageProcessor.generatePreviewImage(src, { longEdge: 400 });
+    expect(await storage.exists(key)).toBe(true);
+    if (storage.kind() === 'local') {
+      const meta = await sharp(storage.resolveLocalPath(key)).metadata();
+      // 800x600 → fit:'inside' inside 400×400 → 400×300.
+      expect(meta.width).toBe(400);
+      expect(meta.height).toBe(300);
+    }
+  });
+
+  test('isPreviewValid returns true for a real preview and false for a missing key', async () => {
+    const src = await makeSourceJpeg(tmpDir, 'preview-valid.jpg');
+    const key = await imageProcessor.generatePreviewImage(src);
+    expect(await imageProcessor.isPreviewValid(key)).toBe(true);
+    expect(await imageProcessor.isPreviewValid('previews/does-not-exist.jpg')).toBe(false);
+  });
+
+  test('isThumbnailValid returns true for a good thumbnail and false for nothing', async () => {
+    const src = await makeSourceJpeg(tmpDir, 'valid-check.jpg');
+    const key = await imageProcessor.generateThumbnail(src);
+    expect(await imageProcessor.isThumbnailValid(key)).toBe(true);
+    expect(await imageProcessor.isThumbnailValid('thumbnails/does-not-exist.jpg')).toBe(false);
+  });
+
+  test('generateVideoPlaceholder writes a thumbnail entirely from buffer', async () => {
+    const key = await imageProcessor.generateVideoPlaceholder('demo.mp4');
+    expect(key).toBe('thumbnails/thumb_demo.jpg');
+    expect(await storage.exists(key)).toBe(true);
+  });
+
+  test('withLocalCopy yields a usable local path on both backends', async () => {
+    const sourceKey = 'fixture/withlocal.jpg';
+    const src = await makeSourceJpeg(tmpDir, 'withlocal.jpg');
+    const buf = await fs.readFile(src);
+    await storage.put(sourceKey, buf, { contentType: 'image/jpeg' });
+
+    const seenSize = await imageProcessor.withLocalCopy(sourceKey, async (localPath) => {
+      const meta = await sharp(localPath).metadata();
+      return meta.width;
+    });
+    expect(seenSize).toBe(800);
+  });
+});

backend/__tests__/integration/incomingInvoiceRebill.test.js

@@ -0,0 +1,211 @@
+/**
+ * Incoming-invoice categorisation + re-bill chain (expenseService) against a
+ * real SQLite schema. Covers the bits unit tests can't: the disposition state
+ * machine, re-categorisation unwind, the per-event PENDING pool + bundling, and
+ * the monthly accumulator immediate-bill — i.e. that categorizeInbound /
+ * billPendingRebills actually mint / amend invoice rows correctly.
+ *
+ * No date-range comparisons are exercised here, so it's safe on SQLite (the
+ * usual PG-vs-SQLite date pitfall — [[feedback_pg_date_columns_serialize]] —
+ * doesn't apply to this path).
+ */
+const { bootCrmDb, seedMinimal } = require('./helpers/crmDb');
+
+// Service-level CRM calls cold-require heavy modules (pdfService, nodemailer)
+// on first use; bump the budget for this file.
+jest.setTimeout(120000);
+
+describe('incoming-invoice categorise / re-bill chain', () => {
+  let db;
+  let cleanup;
+  let adminId;
+  let expenseService;
+
+  beforeAll(async () => {
+    ({ db, cleanup } = await bootCrmDb());
+    // logActivity writes to activity_logs via the GLOBAL db. createInvoice (and
+    // appendToMonthlyDraft) call it INSIDE the transaction we pass them, and a
+    // second write connection deadlocks against the held write lock on
+    // SQLite. It's fire-and-forget audit noise, irrelevant to these
+    // assertions, so stub it BEFORE the services destructure it at require
+    // time. (Production runs Postgres, where the concurrent write is fine.)
+    const dbModule = require('../../src/database/db');
+    dbModule.logActivity = async () => {};
+    ({ adminId } = await seedMinimal(db));
+    expenseService = require('../../src/services/expenseService');
+  }, 120000);
+
+  afterAll(async () => {
+    if (cleanup) await cleanup();
+  });
+
+  const unwrapId = (ins) => (typeof ins[0] === 'object' ? ins[0].id : ins[0]);
+
+  async function captureDoc(overrides = {}) {
+    const ins = await db('inbound_documents').insert({
+      source: 'upload',
+      status: 'unsorted',
+      parse_status: 'pending',
+      parse_method: 'none',
+      supplier_name: 'ACME AG',
+      currency: 'CHF',
+      total_amount_minor: 10000,
+      invoice_date: '2026-06-01',
+      created_at: new Date(),
+      updated_at: new Date(),
+      ...overrides,
+    }).returning('id');
+    return unwrapId(ins);
+  }
+
+  let customerSeq = 0;
+  async function makeCustomer(billingCadence) {
+    customerSeq += 1;
+    const ins = await db('customer_accounts').insert({
+      email: `rebill-${billingCadence || 'event'}-${customerSeq}@example.com`,
+      display_name: `Rebill ${billingCadence || 'event'} ${customerSeq}`,
+      password_hash: 'x',
+      preferred_language: 'de',
+      is_active: 1,
+      billing_cadence: billingCadence || null,
+      created_at: new Date(),
+    }).returning('id');
+    return unwrapId(ins);
+  }
+
+  it('company expense (eigener_aufwand) categorises with no invoice + no customer', async () => {
+    const id = await captureDoc();
+    const doc = await expenseService.categorizeInbound(id, { disposition: 'eigener_aufwand', categoryId: null }, adminId);
+    expect(doc.disposition).toBe('eigener_aufwand');
+    expect(doc.status).toBe('categorized');
+    expect(doc.billedInvoiceId).toBeNull();
+    expect(doc.customerAccountId).toBeNull();
+  });
+
+  it('rebill REQUIRES a customer', async () => {
+    const id = await captureDoc();
+    await expect(expenseService.categorizeInbound(id, { disposition: 'rebill' }, adminId))
+      .rejects.toMatchObject({ code: 'CUSTOMER_REQUIRED' });
+  });
+
+  it('per-event rebill stays PENDING (customer + markup stored, no invoice yet)', async () => {
+    const customerId = await makeCustomer('per_event');
+    const id = await captureDoc({ total_amount_minor: 10000 });
+    const doc = await expenseService.categorizeInbound(id, {
+      disposition: 'rebill', customerAccountId: customerId,
+      markupType: 'percent', markupPercent: 10,
+    }, adminId);
+    expect(doc.disposition).toBe('rebill');
+    expect(doc.customerAccountId).toBe(customerId);
+    expect(doc.billedInvoiceId).toBeNull(); // pending — not billed until bundled
+    expect(doc.markupType).toBe('percent');
+    expect(Number(doc.markupPercent)).toBe(10);
+  });
+
+  it('passthrough never carries a markup, even if one is sent', async () => {
+    const customerId = await makeCustomer('per_event');
+    const id = await captureDoc();
+    const doc = await expenseService.categorizeInbound(id, {
+      disposition: 'durchlaufend', customerAccountId: customerId,
+      markupType: 'percent', markupPercent: 25, // should be ignored
+    }, adminId);
+    expect(doc.disposition).toBe('durchlaufend');
+    expect(doc.customerAccountId).toBe(customerId);
+    expect(doc.markupType).toBe('none');
+    expect(doc.markupPercent).toBeNull();
+    expect(doc.billedInvoiceId).toBeNull();
+  });
+
+  it('billPendingRebills refuses monthly/manual customers (they auto-consolidate)', async () => {
+    const customerId = await makeCustomer('monthly');
+    await expect(expenseService.billPendingRebills(customerId, adminId))
+      .rejects.toMatchObject({ code: 'CADENCE_MISMATCH' });
+  });
+
+  // ── The actual invoice-MINTING paths (billPendingRebills bundling a per-event
+  // customer's pool; monthly-customer immediate-bill onto the running draft)
+  // both call invoiceService.createInvoice INSIDE a db.transaction. createInvoice
+  // claims its sequence number via the global db, which DEADLOCKS against the
+  // held write lock on a SQLite-backed harness (a second write connection blocks
+  // — verified). Production runs Postgres where the concurrent write is fine, so
+  // this is a harness limitation, not a product bug. The line-amount math is
+  // covered by the buildInboundLineItem unit tests, and createInvoice itself by
+  // discountLineItems.test.js. Below we test the UNWIND path against a
+  // hand-crafted billed state so we don't have to mint through createInvoice. ──
+
+  // Build a billed state directly: an invoice with two lines, with the inbound
+  // doc stamped onto the first line as a prior re-bill.
+  async function makeBilledDoc(customerId, { status = 'scheduled', scheduledSendAt = null, isMonthlyDraft = false } = {}) {
+    const invIns = await db('invoices').insert({
+      invoice_number: `R-TEST-${customerSeq}-${Math.floor(Math.random() * 1e9)}`,
+      customer_account_id: customerId,
+      status,
+      scheduled_send_at: scheduledSendAt,
+      is_monthly_draft: isMonthlyDraft,
+      currency: 'CHF',
+      issue_date: '2026-06-01',
+      due_date: '2026-07-01',
+      vat_rate: 0,
+      net_amount_minor: 7000, // 4000 (rebill line) + 3000 (sibling)
+      vat_amount_minor: 0,
+      total_amount_minor: 7000,
+      created_at: new Date(),
+      updated_at: new Date(),
+    }).returning('id');
+    const invoiceId = unwrapId(invIns);
+    const rebillLineIns = await db('invoice_line_items').insert({
+      invoice_id: invoiceId, position: 1, quantity: 1, description: 'Rebill Co (Weiterverrechnung)',
+      unit_price_minor: 4000, discount_percent: 0, line_total_minor: 4000,
+    }).returning('id');
+    const rebillLineId = unwrapId(rebillLineIns);
+    await db('invoice_line_items').insert({
+      invoice_id: invoiceId, position: 2, quantity: 1, description: 'Other line',
+      unit_price_minor: 3000, discount_percent: 0, line_total_minor: 3000,
+    });
+    const id = await captureDoc({ total_amount_minor: 4000, supplier_name: 'Rebill Co' });
+    await db('inbound_documents').where({ id }).update({
+      disposition: 'rebill', status: 'categorized', customer_account_id: customerId,
+      billed_invoice_id: invoiceId, billed_invoice_line_item_id: rebillLineId,
+    });
+    return { id, invoiceId, rebillLineId };
+  }
+
+  it('re-categorising a billed doc UNWINDS its re-bill line + recomputes the (mutable) invoice', async () => {
+    const customerId = await makeCustomer('per_event');
+    const { id, invoiceId, rebillLineId } = await makeBilledDoc(customerId); // scheduled, no send-at → mutable
+
+    const recat = await expenseService.categorizeInbound(id, { disposition: 'eigener_aufwand', categoryId: null }, adminId);
+    expect(recat.disposition).toBe('eigener_aufwand');
+    expect(recat.billedInvoiceId).toBeNull();
+    expect(recat.customerAccountId).toBeNull();
+
+    // The re-bill line is gone; the sibling line remains and net recomputes.
+    expect(await db('invoice_line_items').where({ id: rebillLineId }).first()).toBeUndefined();
+    const after = await db('invoices').where({ id: invoiceId }).first();
+    expect(Number(after.net_amount_minor)).toBe(3000);
+  });
+
+  it('re-categorising a doc billed on an ISSUED invoice is refused (Storno required)', async () => {
+    const customerId = await makeCustomer('per_event');
+    const { id, rebillLineId } = await makeBilledDoc(customerId, { status: 'sent' });
+
+    await expect(expenseService.categorizeInbound(id, { disposition: 'eigener_aufwand' }, adminId))
+      .rejects.toMatchObject({ code: 'INVOICE_LOCKED' });
+    // Nothing was touched — the line survives.
+    expect(await db('invoice_line_items').where({ id: rebillLineId }).first()).toBeDefined();
+  });
+
+  it('re-categorisation moves a pending item between dispositions without a stray invoice', async () => {
+    const customerId = await makeCustomer('per_event');
+    const id = await captureDoc();
+    // passthrough → pending
+    let doc = await expenseService.categorizeInbound(id, { disposition: 'durchlaufend', customerAccountId: customerId }, adminId);
+    expect(doc.customerAccountId).toBe(customerId);
+    expect(doc.billedInvoiceId).toBeNull();
+    // → company expense: customer cleared, still no invoice
+    doc = await expenseService.categorizeInbound(id, { disposition: 'eigener_aufwand' }, adminId);
+    expect(doc.disposition).toBe('eigener_aufwand');
+    expect(doc.customerAccountId).toBeNull();
+    expect(doc.billedInvoiceId).toBeNull();
+  });
+});

backend/__tests__/integration/installFromBackupBoot.test.js

@@ -0,0 +1,214 @@
+/**
+ * Install-from-backup boot hook — pins the trigger-file convention.
+ *
+ * The hook itself depends on `restoreService.restore`, which is hard
+ * to fully exercise in an integration test without a real PG cluster
+ * (sequence resync, DROP/CREATE, etc.). So we stub the actual restore
+ * and verify the BOOT HOOK logic:
+ *
+ *   - No trigger file → no-op, ran=false
+ *   - Empty trigger file → picks newest manifest from manifests/
+ *   - Non-empty trigger file → uses the path inside
+ *   - DB not empty → refuses (no restore call)
+ *   - DB not empty + FORCE env → proceeds
+ *   - Successful restore → deletes trigger file
+ *   - Failed restore → leaves trigger file in place
+ *
+ * These are the surfaces an admin will hit when actually using the
+ * feature — the docker-compose-on-real-PG end-to-end test belongs in
+ * the follow-up CI work captured as task #7 earlier today.
+ */
+
+const fs = require('fs');
+const path = require('path');
+
+const { bootCrmDb } = require('./helpers/crmDb');
+
+// Stub the heavy lifting so the test stays fast + portable.
+const mockRestore = jest.fn();
+jest.mock('../../src/services/restoreService', () => ({
+  restoreService: {
+    restore: (...args) => mockRestore(...args),
+  },
+}));
+
+jest.setTimeout(120000);
+
+describe('installFromBackupBoot', () => {
+  let db;
+  let cleanup;
+  let storagePath;
+  let backupRoot;
+  let manifestsDir;
+  let tryInstallFromBackup;
+  let originalBackupRootEnv;
+  let originalForceEnv;
+
+  beforeAll(async () => {
+    ({ db, cleanup } = await bootCrmDb());
+    storagePath = process.env.STORAGE_PATH;
+    backupRoot = path.join(storagePath, 'backup');
+    manifestsDir = path.join(backupRoot, 'manifests');
+    fs.mkdirSync(manifestsDir, { recursive: true });
+
+    originalBackupRootEnv = process.env.BACKUP_ROOT;
+    originalForceEnv = process.env.INSTALL_FROM_BACKUP_FORCE;
+    process.env.BACKUP_ROOT = backupRoot;
+
+    ({ tryInstallFromBackup } = require('../../src/services/_installFromBackupBoot'));
+  }, 120000);
+
+  afterAll(async () => {
+    if (originalBackupRootEnv === undefined) {
+      delete process.env.BACKUP_ROOT;
+    } else {
+      process.env.BACKUP_ROOT = originalBackupRootEnv;
+    }
+    if (originalForceEnv === undefined) {
+      delete process.env.INSTALL_FROM_BACKUP_FORCE;
+    } else {
+      process.env.INSTALL_FROM_BACKUP_FORCE = originalForceEnv;
+    }
+    if (cleanup) await cleanup();
+  });
+
+  beforeEach(async () => {
+    mockRestore.mockReset();
+    mockRestore.mockResolvedValue({ success: true });
+    delete process.env.INSTALL_FROM_BACKUP_FORCE;
+
+    // Clean trigger files + manifests between tests
+    for (const name of ['RESTORE_ON_INSTALL', 'RESTORE_ON_INSTALL.txt']) {
+      const p = path.join(backupRoot, name);
+      if (fs.existsSync(p)) fs.unlinkSync(p);
+    }
+    for (const f of fs.readdirSync(manifestsDir)) {
+      fs.unlinkSync(path.join(manifestsDir, f));
+    }
+
+    // Reset DB to fresh-install state
+    await db('events').del();
+    // Leave admin_users alone — fresh-install state has 1 row.
+  });
+
+  it('no trigger file → no-op', async () => {
+    const result = await tryInstallFromBackup(db);
+    expect(result.ran).toBe(false);
+    expect(mockRestore).not.toHaveBeenCalled();
+  });
+
+  it('empty trigger file picks the newest manifest from manifests/', async () => {
+    const older = path.join(manifestsDir, 'backup-manifest-001.json');
+    const newer = path.join(manifestsDir, 'backup-manifest-002.json');
+    fs.writeFileSync(older, '{}');
+    // Set the newer file's mtime slightly later so it wins the sort
+    const past = new Date(Date.now() - 60_000);
+    fs.utimesSync(older, past, past);
+    fs.writeFileSync(newer, '{}');
+
+    // Empty trigger
+    fs.writeFileSync(path.join(backupRoot, 'RESTORE_ON_INSTALL'), '');
+
+    const result = await tryInstallFromBackup(db);
+    expect(result.ran).toBe(true);
+    expect(result.manifestPath).toBe(newer);
+    expect(mockRestore).toHaveBeenCalledWith(expect.objectContaining({
+      source: 'local',
+      manifestPath: newer,
+      restoreType: 'full',
+      force: true,
+      skipPreBackup: true,
+    }));
+  });
+
+  it('non-empty trigger file uses the path inside', async () => {
+    const specific = path.join(manifestsDir, 'backup-manifest-specific.json');
+    fs.writeFileSync(specific, '{}');
+
+    // Relative to backupRoot
+    fs.writeFileSync(
+      path.join(backupRoot, 'RESTORE_ON_INSTALL'),
+      'manifests/backup-manifest-specific.json\n',
+    );
+
+    const result = await tryInstallFromBackup(db);
+    expect(result.ran).toBe(true);
+    expect(result.manifestPath).toBe(specific);
+  });
+
+  it('deletes the trigger file after a successful restore', async () => {
+    const manifest = path.join(manifestsDir, 'backup-manifest-001.json');
+    fs.writeFileSync(manifest, '{}');
+    const triggerPath = path.join(backupRoot, 'RESTORE_ON_INSTALL');
+    fs.writeFileSync(triggerPath, '');
+
+    await tryInstallFromBackup(db);
+    expect(fs.existsSync(triggerPath)).toBe(false);
+  });
+
+  it('leaves the trigger file in place when restore throws', async () => {
+    mockRestore.mockRejectedValueOnce(new Error('restore exploded'));
+    const manifest = path.join(manifestsDir, 'backup-manifest-001.json');
+    fs.writeFileSync(manifest, '{}');
+    const triggerPath = path.join(backupRoot, 'RESTORE_ON_INSTALL');
+    fs.writeFileSync(triggerPath, '');
+
+    const result = await tryInstallFromBackup(db);
+    expect(result.ran).toBe(false);
+    expect(result.error).toMatch(/restore exploded/);
+    expect(fs.existsSync(triggerPath)).toBe(true);
+  });
+
+  it('refuses to run when the install already has events', async () => {
+    // Simulate an install with existing data
+    await db('events').insert({
+      slug: 'existing-event',
+      event_name: 'Existing Event',
+      event_type: 'wedding',
+      event_date: new Date(),
+      host_email: 'host@example.com',
+      admin_email: 'host@example.com',
+      expires_at: new Date(Date.now() + 7 * 24 * 60 * 60 * 1000),
+      share_link: 'existing-event-token',
+      password_hash: 'dummy-hash-for-test',
+      created_at: new Date(),
+    });
+
+    const manifest = path.join(manifestsDir, 'backup-manifest-001.json');
+    fs.writeFileSync(manifest, '{}');
+    fs.writeFileSync(path.join(backupRoot, 'RESTORE_ON_INSTALL'), '');
+
+    const result = await tryInstallFromBackup(db);
+    expect(result.ran).toBe(false);
+    expect(result.error).toMatch(/Database not empty/);
+    expect(mockRestore).not.toHaveBeenCalled();
+
+    // Trigger file should NOT be deleted — admin needs to fix + retry
+    expect(fs.existsSync(path.join(backupRoot, 'RESTORE_ON_INSTALL'))).toBe(true);
+  });
+
+  it('proceeds when INSTALL_FROM_BACKUP_FORCE=true even with existing data', async () => {
+    await db('events').insert({
+      slug: 'existing-event-2',
+      event_name: 'Existing Event 2',
+      event_type: 'wedding',
+      event_date: new Date(),
+      host_email: 'host@example.com',
+      admin_email: 'host@example.com',
+      expires_at: new Date(Date.now() + 7 * 24 * 60 * 60 * 1000),
+      share_link: 'existing-event-2-token',
+      password_hash: 'dummy-hash-for-test-2',
+      created_at: new Date(),
+    });
+
+    const manifest = path.join(manifestsDir, 'backup-manifest-001.json');
+    fs.writeFileSync(manifest, '{}');
+    fs.writeFileSync(path.join(backupRoot, 'RESTORE_ON_INSTALL'), '');
+
+    process.env.INSTALL_FROM_BACKUP_FORCE = 'true';
+    const result = await tryInstallFromBackup(db);
+
+    expect(result.ran).toBe(true);
+    expect(mockRestore).toHaveBeenCalled();
+  });
+});

backend/__tests__/integration/invoiceDunning.test.js

@@ -0,0 +1,115 @@
+/**
+ * Dunning / Mahngebühr logic — the tax-sensitive bits added in the dunning
+ * rework. Covers the fee math (flat / percent), the VAT toggle gating
+ * (incl. the "no-op when the org has no VAT rate" requirement), per-reminder
+ * accumulation (2nd = 1×, 3rd = 2×), invoice immutability (the fee never
+ * changes the issued invoice total), and the 3-reminder cap.
+ *
+ * The Mahnung PDF render is stubbed — PDF rendering (fonts) is flaky in CI and
+ * is verified manually; here we assert the data/immutability behaviour.
+ */
+const { bootCrmDb, seedMinimal } = require('./helpers/crmDb');
+
+// bootCrmDb runs the full core-migration set in beforeAll; under full-suite
+// parallel load on a small CI runner that can exceed the 5s default. Match the
+// other migration-heavy CRM suites (discountLineItems, incomingInvoiceRebill).
+jest.setTimeout(120000);
+
+let db;
+let cleanup;
+let invoiceService;
+let ids;
+
+async function setSetting(key, value) {
+  const { upsertAppSetting } = require('../../src/utils/appSettings');
+  await upsertAppSetting(key, JSON.stringify(value), 'crm');
+}
+
+beforeAll(async () => {
+  ({ db, cleanup } = await bootCrmDb());
+  ids = await seedMinimal(db);
+  try { await db('customer_accounts').where({ id: ids.customerId }).update({ feature_bills: true }); } catch (_) {}
+  invoiceService = require('../../src/services/invoiceService');
+  // Stub the (flaky) PDF render so applyReminder exercises its data path.
+  // eslint-disable-next-line global-require
+  const pdfService = require('../../src/services/pdfService');
+  pdfService.renderInvoiceToBuffer = async () => Buffer.from('%PDF-stub');
+});
+
+afterAll(async () => { await cleanup(); });
+
+describe('dunning fee resolvers', () => {
+  test('flat fee, no VAT', async () => {
+    await setSetting('crm_invoices_late_fee_enabled', true);
+    await setSetting('crm_invoices_late_fee_type', 'flat');
+    await setSetting('crm_invoices_late_fee_minor', 2000);
+    await setSetting('crm_invoices_late_fee_vat_enabled', false);
+    const inv = { total_amount_minor: 100000 };
+    expect(await invoiceService.resolveLateFeeNetMinor(inv)).toBe(2000);
+    expect(await invoiceService.resolveLateFeeVatRate()).toBe(0);
+    expect(await invoiceService.resolvePerReminderFeeMinor(inv)).toBe(2000);
+  });
+
+  test('percent fee = % of the invoice gross', async () => {
+    await setSetting('crm_invoices_late_fee_type', 'percent');
+    await setSetting('crm_invoices_late_fee_percent', 5);
+    expect(await invoiceService.resolveLateFeeNetMinor({ total_amount_minor: 100000 })).toBe(5000);
+  });
+
+  test('VAT toggle applies the org rate, but is a NO-OP when the org has no VAT rate', async () => {
+    await setSetting('crm_invoices_late_fee_type', 'flat');
+    await setSetting('crm_invoices_late_fee_minor', 2000);
+    await setSetting('crm_invoices_late_fee_vat_enabled', true);
+
+    await db('business_profile').where({ id: 1 }).update({ vat_rate_default: 8.1 });
+    expect(await invoiceService.resolveLateFeeVatRate()).toBeCloseTo(8.1);
+    expect(await invoiceService.resolvePerReminderFeeMinor({ total_amount_minor: 0 }))
+      .toBe(2000 + Math.round(2000 * 8.1 / 100)); // net + VAT
+
+    // Org doesn't charge VAT → toggle adds nothing (Mara's requirement).
+    await db('business_profile').where({ id: 1 }).update({ vat_rate_default: 0 });
+    expect(await invoiceService.resolveLateFeeVatRate()).toBe(0);
+    expect(await invoiceService.resolvePerReminderFeeMinor({ total_amount_minor: 0 })).toBe(2000);
+  });
+});
+
+describe('applyReminder — dunning-document model', () => {
+  let invoiceId;
+  let originalTotal;
+
+  beforeAll(async () => {
+    await setSetting('crm_invoices_late_fee_enabled', true);
+    await setSetting('crm_invoices_late_fee_type', 'flat');
+    await setSetting('crm_invoices_late_fee_minor', 2000);
+    await setSetting('crm_invoices_late_fee_vat_enabled', false);
+    const res = await invoiceService.createInvoice({
+      customerAccountId: ids.customerId,
+      currency: 'CHF',
+      vatRate: 0,
+      lineItems: [{ description: 'Service', quantity: 1, unit_price_minor: 100000 }],
+    }, ids.adminId);
+    invoiceId = res.invoiceIds[0];
+    originalTotal = Number((await db('invoices').where({ id: invoiceId }).first()).total_amount_minor);
+  });
+
+  test('level 2 tracks one fee and leaves the invoice total immutable', async () => {
+    const data = await invoiceService.getInvoiceById(invoiceId);
+    await invoiceService.applyReminder(data.invoice, data.lineItems, 2, ids.adminId);
+    const inv = await db('invoices').where({ id: invoiceId }).first();
+    expect(inv.reminder_level).toBe(2);
+    expect(Number(inv.late_fee_amount_minor)).toBe(2000);
+    expect(Number(inv.total_amount_minor)).toBe(originalTotal); // never mutated
+  });
+
+  test('level 3 accumulates the fee to 2×, total still immutable', async () => {
+    const data = await invoiceService.getInvoiceById(invoiceId);
+    await invoiceService.applyReminder(data.invoice, data.lineItems, 3, ids.adminId);
+    const inv = await db('invoices').where({ id: invoiceId }).first();
+    expect(Number(inv.late_fee_amount_minor)).toBe(4000);
+    expect(Number(inv.total_amount_minor)).toBe(originalTotal);
+  });
+
+  test('sendReminder refuses to exceed level 3', async () => {
+    await expect(invoiceService.sendReminder(invoiceId, 4, ids.adminId)).rejects.toThrow();
+  });
+});

backend/__tests__/integration/maintenanceJobState.test.js

@@ -0,0 +1,233 @@
+/**
+ * Shared run state for the maintenance sweeps (#1181).
+ *
+ * The behaviour that matters here cannot be observed from one process holding
+ * a module-level flag, which is exactly why the flag moved into the database.
+ * A second replica is simulated the only way that is honest in a single-process
+ * test: by asserting on the shared row itself, and by driving claim() twice —
+ * a second caller getting null is precisely what a second replica gets.
+ */
+
+const fs = require('fs');
+const path = require('path');
+const os = require('os');
+const express = require('express');
+const request = require('supertest');
+
+describe('maintenance job state (#1181)', () => {
+  let tmpDir; let db; let app; let jobs;
+
+  const dimStatus = () => request(app).get('/api/admin/photos/repair-dimensions/status');
+  const capStatus = () => request(app).get('/api/admin/photos/repair-capture-dates/status');
+
+  beforeAll(async () => {
+    tmpDir = await fs.promises.mkdtemp(path.join(os.tmpdir(), 'picpeak-mjs-'));
+    process.env.NODE_ENV = 'test';
+    process.env.TEST_DATABASE_PATH = path.join(tmpDir, 'data', 'db.sqlite');
+    await fs.promises.mkdir(path.dirname(process.env.TEST_DATABASE_PATH), { recursive: true });
+    process.env.STORAGE_PATH = path.join(tmpDir, 'storage');
+    process.env.JWT_SECRET = process.env.JWT_SECRET || 'mjs-secret';
+
+    jest.resetModules();
+    jest.doMock('../../src/middleware/auth', () => ({
+      adminAuth: (req, _res, next) => { req.admin = { id: 1, username: 'tester', roleName: 'admin' }; next(); },
+    }));
+    jest.doMock('../../src/middleware/permissions', () => ({
+      requirePermission: () => (_req, _res, next) => next(),
+    }));
+    jest.doMock('../../src/utils/logger', () => ({
+      debug: jest.fn(), info: jest.fn(), warn: jest.fn(), error: jest.fn(),
+    }));
+
+    ({ db } = await require('./helpers/crmDb').bootCrmDb());
+    jobs = require('../../src/services/maintenanceJobState');
+
+    app = express();
+    app.use(express.json());
+    app.use('/api/admin/photos', require('../../src/routes/adminPhotoDimensions'));
+  }, 180000);
+
+  afterAll(async () => {
+    if (db) await db.destroy?.();
+    await fs.promises.rm(tmpDir, { recursive: true, force: true }).catch(() => {});
+  });
+
+  beforeEach(async () => {
+    await db('maintenance_jobs').update({
+      is_running: false, started_at: null, heartbeat_at: null, finished_at: null, last_result: null, owner: null, claim_token: null,
+    });
+  });
+
+  test('the lease table is kept out of .picpeak archives', () => {
+    // It is live state, not data. An archive taken mid-sweep would otherwise
+    // carry is_running = true and a claim token owned by a process on the
+    // SOURCE install; restored inside the staleness window, the target reports
+    // the job as running and refuses new POSTs with no runner to release it.
+    // The importer filters on this same set, so archives written before the
+    // exclusion are skipped on restore too.
+    const { EXCLUDED_TABLES } = require('../../src/services/picpeakExportService');
+    expect(EXCLUDED_TABLES.has('maintenance_jobs')).toBe(true);
+  });
+
+  test('the migration seeds a row for each job', async () => {
+    const names = await db('maintenance_jobs').pluck('job_name');
+    expect(names.sort()).toEqual(['photo_capture_date_backfill', 'photo_dimension_repair']);
+  });
+
+  test('a second claim is refused while the first is alive', async () => {
+    expect(await jobs.claim(jobs.JOB_DIMENSION_REPAIR)).toEqual(expect.any(String));
+    // What a second replica's POST does. Nothing about the first claim lives in
+    // this process, so this is the same question the other replica asks.
+    expect(await jobs.claim(jobs.JOB_DIMENSION_REPAIR)).toBeNull();
+  });
+
+  test('the two jobs claim independently', async () => {
+    expect(await jobs.claim(jobs.JOB_DIMENSION_REPAIR)).toEqual(expect.any(String));
+    expect(await jobs.claim(jobs.JOB_CAPTURE_DATE_BACKFILL)).toEqual(expect.any(String));
+  });
+
+  test('each claim gets a distinct token', async () => {
+    const first = await jobs.claim(jobs.JOB_DIMENSION_REPAIR);
+    await jobs.release(jobs.JOB_DIMENSION_REPAIR, first);
+    const second = await jobs.claim(jobs.JOB_DIMENSION_REPAIR);
+    // Same process, same pid — so an owner string would have collided here and
+    // the fencing below would be worthless.
+    expect(second).not.toBe(first);
+  });
+
+  test('a claim whose heartbeat has gone quiet can be taken over', async () => {
+    expect(await jobs.claim(jobs.JOB_DIMENSION_REPAIR)).toEqual(expect.any(String));
+    expect(await jobs.claim(jobs.JOB_DIMENSION_REPAIR)).toBeNull();
+
+    // The replica holding it was killed: no release, no further heartbeats.
+    const longAgo = new Date(Date.now() - jobs.DEFAULT_STALE_MS - 60000).toISOString();
+    await db('maintenance_jobs').where({ job_name: jobs.JOB_DIMENSION_REPAIR }).update({ heartbeat_at: longAgo });
+
+    expect(await jobs.claim(jobs.JOB_DIMENSION_REPAIR)).toEqual(expect.any(String));
+  });
+
+  test('a superseded runner cannot renew its lease', async () => {
+    const oldToken = await jobs.claim(jobs.JOB_DIMENSION_REPAIR);
+    const longAgo = new Date(Date.now() - jobs.DEFAULT_STALE_MS - 60000).toISOString();
+    await db('maintenance_jobs').where({ job_name: jobs.JOB_DIMENSION_REPAIR }).update({ heartbeat_at: longAgo });
+    const newToken = await jobs.claim(jobs.JOB_DIMENSION_REPAIR);
+    expect(newToken).toEqual(expect.any(String));
+
+    // The old runner is still alive and mid-loop. Its renewal must tell it so,
+    // which is what makes the route loop stop instead of running alongside the
+    // new owner.
+    expect(await jobs.heartbeat(jobs.JOB_DIMENSION_REPAIR, oldToken)).toBe(false);
+    expect(await jobs.heartbeat(jobs.JOB_DIMENSION_REPAIR, newToken)).toBe(true);
+  });
+
+  test('a superseded runner cannot release the new owner\'s claim', async () => {
+    const oldToken = await jobs.claim(jobs.JOB_CAPTURE_DATE_BACKFILL);
+    const longAgo = new Date(Date.now() - jobs.DEFAULT_STALE_MS - 60000).toISOString();
+    await db('maintenance_jobs').where({ job_name: jobs.JOB_CAPTURE_DATE_BACKFILL }).update({ heartbeat_at: longAgo });
+    const newToken = await jobs.claim(jobs.JOB_CAPTURE_DATE_BACKFILL);
+
+    // The old runner finishes late and tries to write its result. Unfenced,
+    // this cleared is_running under the new owner and let a THIRD sweep start.
+    expect(await jobs.release(jobs.JOB_CAPTURE_DATE_BACKFILL, oldToken, { success: 999, noExif: 0, failed: 0 })).toBe(false);
+
+    const state = await jobs.read(jobs.JOB_CAPTURE_DATE_BACKFILL);
+    expect(state.isRunning).toBe(true);
+    expect(state.lastResult).toBeNull();
+    // And the row is still the new owner's to release.
+    expect(await jobs.release(jobs.JOB_CAPTURE_DATE_BACKFILL, newToken, { success: 1, noExif: 0, failed: 0 })).toBe(true);
+  });
+
+  test('a stale run reads as not running, so the button comes back', async () => {
+    await jobs.claim(jobs.JOB_DIMENSION_REPAIR);
+    expect((await jobs.read(jobs.JOB_DIMENSION_REPAIR)).isRunning).toBe(true);
+
+    const longAgo = new Date(Date.now() - jobs.DEFAULT_STALE_MS - 60000).toISOString();
+    await db('maintenance_jobs').where({ job_name: jobs.JOB_DIMENSION_REPAIR }).update({ heartbeat_at: longAgo });
+
+    // is_running is still true in the row — nothing released it — but a status
+    // poll must not leave the operator staring at a job that cannot finish.
+    expect((await db('maintenance_jobs').where({ job_name: jobs.JOB_DIMENSION_REPAIR }).first()).is_running).toBeTruthy();
+    expect((await jobs.read(jobs.JOB_DIMENSION_REPAIR)).isRunning).toBe(false);
+  });
+
+  test('a heartbeat keeps a long run claimed', async () => {
+    const token = await jobs.claim(jobs.JOB_DIMENSION_REPAIR);
+    const longAgo = new Date(Date.now() - jobs.DEFAULT_STALE_MS - 60000).toISOString();
+    await db('maintenance_jobs').where({ job_name: jobs.JOB_DIMENSION_REPAIR }).update({ heartbeat_at: longAgo });
+
+    expect(await jobs.heartbeat(jobs.JOB_DIMENSION_REPAIR, token)).toBe(true);
+
+    expect(await jobs.claim(jobs.JOB_DIMENSION_REPAIR)).toBeNull();
+    expect((await jobs.read(jobs.JOB_DIMENSION_REPAIR)).isRunning).toBe(true);
+  });
+
+  test('release stores the result and read gives it back parsed', async () => {
+    const token = await jobs.claim(jobs.JOB_CAPTURE_DATE_BACKFILL);
+    await jobs.release(jobs.JOB_CAPTURE_DATE_BACKFILL, token, { success: 3, noExif: 2, failed: 1 });
+
+    const state = await jobs.read(jobs.JOB_CAPTURE_DATE_BACKFILL);
+    expect(state.isRunning).toBe(false);
+    expect(state.lastResult).toEqual({ success: 3, noExif: 2, failed: 1 });
+  });
+
+  test('releasing without a result keeps the previous run visible', async () => {
+    const first = await jobs.claim(jobs.JOB_CAPTURE_DATE_BACKFILL);
+    await jobs.release(jobs.JOB_CAPTURE_DATE_BACKFILL, first, { success: 7, noExif: 0, failed: 0 });
+
+    // The "nothing to do" path: claimed, found no candidates, released. It must
+    // not blank the numbers the last real run reported.
+    const second = await jobs.claim(jobs.JOB_CAPTURE_DATE_BACKFILL);
+    await jobs.release(jobs.JOB_CAPTURE_DATE_BACKFILL, second);
+
+    expect((await jobs.read(jobs.JOB_CAPTURE_DATE_BACKFILL)).lastResult).toEqual({ success: 7, noExif: 0, failed: 0 });
+  });
+
+  test('a malformed result does not take the status endpoint down', async () => {
+    await db('maintenance_jobs').where({ job_name: jobs.JOB_DIMENSION_REPAIR }).update({ last_result: 'not json' });
+    const state = await jobs.read(jobs.JOB_DIMENSION_REPAIR);
+    expect(state.lastResult).toBeNull();
+    expect(state.isRunning).toBe(false);
+  });
+
+  test('both status endpoints report the shared row, not process memory', async () => {
+    await jobs.claim(jobs.JOB_DIMENSION_REPAIR);
+    const capToken = await jobs.claim(jobs.JOB_CAPTURE_DATE_BACKFILL);
+    await jobs.release(jobs.JOB_CAPTURE_DATE_BACKFILL, capToken, { success: 1, noExif: 0, failed: 0 });
+
+    // Written straight to the row, exactly as another replica would have.
+    const dim = await dimStatus();
+    expect(dim.status).toBe(200);
+    expect(dim.body.isRunning).toBe(true);
+
+    const cap = await capStatus();
+    expect(cap.status).toBe(200);
+    expect(cap.body.isRunning).toBe(false);
+    expect(cap.body.lastResult).toEqual({ success: 1, noExif: 0, failed: 0 });
+  });
+
+  test('a POST is refused while another replica holds the claim', async () => {
+    // The claim was taken by "another replica" — this process knows nothing
+    // about it beyond the row.
+    await jobs.claim(jobs.JOB_CAPTURE_DATE_BACKFILL);
+
+    const res = await request(app).post('/api/admin/photos/repair-capture-dates');
+    expect(res.status).toBe(409);
+
+    const dimRes = await request(app).post('/api/admin/photos/repair-dimensions');
+    // The other job is untouched by that claim, so it is free to start.
+    expect(dimRes.status).toBe(200);
+  });
+
+  test('the no-op path releases the claim it took', async () => {
+    // No photos at all, so both endpoints take their "nothing to do" exit.
+    await db('photos').del();
+
+    const res = await request(app).post('/api/admin/photos/repair-capture-dates');
+    expect(res.body.count).toBe(0);
+
+    const row = await db('maintenance_jobs').where({ job_name: jobs.JOB_CAPTURE_DATE_BACKFILL }).first();
+    expect(row.is_running).toBeFalsy();
+    // ...and a second POST is therefore accepted rather than 409ing forever.
+    expect((await request(app).post('/api/admin/photos/repair-capture-dates')).status).toBe(200);
+  });
+});

backend/__tests__/integration/maintenanceJobStatePg.test.js

@@ -0,0 +1,102 @@
+/**
+ * PostgreSQL checks for the shared maintenance-job state (#1181).
+ * Gated: runs only when PICPEAK_PG_TEST_URL points at a throwaway database, e.g.
+ *   PICPEAK_PG_TEST_URL="postgres://picpeak:picpeak_secure_pass_2024@127.0.0.1:7102/picpeak_mjs_test" \
+ *     npx jest __tests__/integration/maintenanceJobStatePg.test.js
+ *
+ * What SQLite cannot answer: the claim leans on comparing a `timestamp` column
+ * against an ISO-8601 string, and on an UPDATE ... WHERE guard being atomic
+ * under real concurrent connections. SQLite compares those strings
+ * lexicographically and serialises writes anyway, so it would pass either way —
+ * exactly the shape of divergence that has bitten this repo before.
+ */
+
+const knex = require('knex');
+
+const PG_URL = process.env.PICPEAK_PG_TEST_URL;
+const maybe = PG_URL ? describe : describe.skip;
+
+maybe('maintenance job state on Postgres', () => {
+  let pgDb;
+  let jobs;
+  const JOB = 'photo_dimension_repair';
+
+  beforeAll(async () => {
+    pgDb = knex({ client: 'pg', connection: PG_URL, pool: { min: 0, max: 10 } });
+    await pgDb.raw('DROP TABLE IF EXISTS maintenance_jobs');
+    await require('../../migrations/core/179_maintenance_job_state').up(pgDb);
+
+    jest.resetModules();
+    jest.doMock('../../src/database/db', () => ({ db: pgDb }));
+    jest.doMock('../../src/utils/logger', () => ({
+      debug: jest.fn(), info: jest.fn(), warn: jest.fn(), error: jest.fn(),
+    }));
+    jobs = require('../../src/services/maintenanceJobState');
+  }, 60000);
+
+  afterAll(async () => {
+    jest.dontMock('../../src/database/db');
+    if (pgDb) await pgDb.destroy();
+  });
+
+  beforeEach(async () => {
+    await pgDb('maintenance_jobs').update({
+      is_running: false, started_at: null, heartbeat_at: null, finished_at: null, last_result: null, owner: null, claim_token: null,
+    });
+  });
+
+  test('the ISO-string cutoff really compares as a timestamp, not as text', async () => {
+    expect(await jobs.claim(JOB)).toEqual(expect.any(String));
+    expect(await jobs.claim(JOB)).toBeNull();
+
+    await pgDb('maintenance_jobs').where({ job_name: JOB })
+      .update({ heartbeat_at: new Date(Date.now() - jobs.DEFAULT_STALE_MS - 60000).toISOString() });
+
+    // If Postgres had rejected or mis-cast the ISO string this would either
+    // throw or never match.
+    expect(await jobs.claim(JOB)).toEqual(expect.any(String));
+
+    const row = await pgDb('maintenance_jobs').where({ job_name: JOB }).first();
+    expect(row.heartbeat_at).toBeInstanceOf(Date);
+  });
+
+  test('concurrent claims on real connections produce exactly one winner', async () => {
+    // The whole point of the conditional UPDATE. Ten connections race; nine
+    // must lose. SQLite cannot demonstrate this — it serialises writers.
+    const results = await Promise.all(Array.from({ length: 10 }, () => jobs.claim(JOB)));
+    expect(results.filter(Boolean)).toHaveLength(1);
+    // ...and the winner holds a token nobody else can forge.
+    expect(results.find(Boolean)).toEqual(expect.any(String));
+  });
+
+  test('a released job can be re-claimed exactly once again', async () => {
+    const token = await jobs.claim(JOB);
+    await jobs.release(JOB, token, { success: 2, failed: 0 });
+
+    const results = await Promise.all(Array.from({ length: 5 }, () => jobs.claim(JOB)));
+    expect(results.filter(Boolean)).toHaveLength(1);
+    expect((await jobs.read(JOB)).lastResult).toEqual({ success: 2, failed: 0 });
+  });
+
+  test('a superseded runner is fenced out on real Postgres', async () => {
+    const oldToken = await jobs.claim(JOB);
+    await pgDb('maintenance_jobs').where({ job_name: JOB })
+      .update({ heartbeat_at: new Date(Date.now() - jobs.DEFAULT_STALE_MS - 60000).toISOString() });
+    const newToken = await jobs.claim(JOB);
+
+    expect(await jobs.heartbeat(JOB, oldToken)).toBe(false);
+    expect(await jobs.release(JOB, oldToken, { success: 999, failed: 0 })).toBe(false);
+    // The new owner still holds it, with its result unwritten.
+    expect((await jobs.read(JOB)).isRunning).toBe(true);
+    expect(await jobs.release(JOB, newToken, { success: 4, failed: 0 })).toBe(true);
+  });
+
+  test('read() reports a live claim as running and a stale one as not', async () => {
+    await jobs.claim(JOB);
+    expect((await jobs.read(JOB)).isRunning).toBe(true);
+
+    await pgDb('maintenance_jobs').where({ job_name: JOB })
+      .update({ heartbeat_at: new Date(Date.now() - jobs.DEFAULT_STALE_MS - 1000).toISOString() });
+    expect((await jobs.read(JOB)).isRunning).toBe(false);
+  });
+});

backend/__tests__/integration/picpeakCrossEngine.test.js

@@ -0,0 +1,197 @@
+'use strict';
+
+/**
+ * Cross-engine .picpeak restore policy (#1041): a SQLite archive restored onto
+ * a PostgreSQL instance — the official small-install → full-stack upgrade
+ * path — now allowed by validateManifest's direction rule instead of the
+ * former CLI-only allowEngineSwitch flag. The coercion engine itself
+ * (typedColumnsFor / epochToIso / coerceForTargetEngine) landed with #1039;
+ * these tests pin the direction policy and the coercion's cross-engine
+ * value-correctness.
+ *
+ * Ungated: validateManifest direction rules and the pure coercion units.
+ * The reverse direction (pg backup onto a sqlite instance) staying blocked is
+ * pinned by picpeakRoundtrip.test.js, which runs on the real sqlite harness.
+ *
+ * Gated on PICPEAK_PG_TEST_URL (same contract as picpeakRestorePg.test.js):
+ * sqlite-shaped NDJSON rows land in real Postgres with correct stored VALUES,
+ * not just row counts, e.g.
+ *   PICPEAK_PG_TEST_URL="postgres://picpeak:pw@127.0.0.1:7102/picpeak_xengine_test" \
+ *     npx jest __tests__/integration/picpeakCrossEngine.test.js
+ */
+const knexLib = require('knex');
+
+describe('validateManifest cross-engine direction (pg target)', () => {
+  let validateManifest;
+
+  beforeAll(() => {
+    jest.resetModules();
+    jest.doMock('../../knexfile', () => ({ client: 'pg' }));
+    // validateManifest wraps its knex_migrations lookup in try/catch — a
+    // throwing stub simply skips the forward-only check, which is not under
+    // test here.
+    jest.doMock('../../src/database/db', () => ({ db: () => { throw new Error('stub'); } }));
+    ({ validateManifest } = require('../../src/services/picpeakImportService'));
+  });
+
+  afterAll(() => {
+    jest.dontMock('../../src/database/db');
+    jest.dontMock('../../knexfile');
+    jest.resetModules();
+  });
+
+  it('allows a sqlite backup onto a pg instance (upgrade direction)', async () => {
+    const blockers = await validateManifest({
+      kind: 'picpeak-backup', format: 1, database: { engine: 'sqlite' }, tables: {},
+    });
+    expect(blockers.filter((b) => /engine/i.test(b))).toHaveLength(0);
+  });
+
+  it('still allows same-engine pg → pg', async () => {
+    const blockers = await validateManifest({
+      kind: 'picpeak-backup', format: 1, database: { engine: 'pg' }, tables: {},
+    });
+    expect(blockers.filter((b) => /engine/i.test(b))).toHaveLength(0);
+  });
+});
+
+describe('epochToIso (landed with #1039)', () => {
+  let epochToIso;
+
+  beforeAll(() => {
+    jest.resetModules();
+    ({ epochToIso } = require('../../src/services/picpeakImportService'));
+  });
+
+  it('converts epoch milliseconds', () => {
+    expect(epochToIso(1723400000000)).toBe('2024-08-11T18:13:20.000Z');
+  });
+
+  it('converts epoch SECONDS to the same instant, not January 1970', () => {
+    expect(epochToIso(1723400000)).toBe('2024-08-11T18:13:20.000Z');
+  });
+
+  it('converts numeric strings', () => {
+    expect(epochToIso('1723400000000')).toBe('2024-08-11T18:13:20.000Z');
+  });
+
+  it('passes non-numeric values through untouched', () => {
+    expect(epochToIso('2026-08-12 10:00:00')).toBe('2026-08-12 10:00:00');
+  });
+});
+
+describe('coerceForTargetEngine on sqlite-shaped rows', () => {
+  let coerceForTargetEngine;
+
+  beforeAll(() => {
+    jest.resetModules();
+    ({ coerceForTargetEngine } = require('../../src/services/picpeakImportService'));
+  });
+
+  const types = { timestamps: ['created_at', 'expires_at'], booleans: ['is_active'] };
+
+  it('coerces 0/1 booleans and epoch timestamps, leaves date strings alone', () => {
+    const [row] = coerceForTargetEngine(
+      [{ id: 1, is_active: 1, created_at: 1723400000000, expires_at: '2026-09-01 12:00:00' }],
+      types
+    );
+    expect(row.is_active).toBe(true);
+    expect(row.created_at).toBe('2024-08-11T18:13:20.000Z');
+    expect(row.expires_at).toBe('2026-09-01 12:00:00'); // pg parses this natively
+  });
+
+  it('coerces falsy variants and passes null/empty through', () => {
+    const [row] = coerceForTargetEngine(
+      [{ is_active: 0, created_at: null, expires_at: '' }],
+      types
+    );
+    expect(row.is_active).toBe(false);
+    expect(row.created_at).toBeNull();
+    expect(row.expires_at).toBe('');
+  });
+});
+
+// ── Real-Postgres integration (gated) ────────────────────────────────────────
+const PG_URL = process.env.PICPEAK_PG_TEST_URL;
+const maybe = PG_URL ? describe : describe.skip;
+
+maybe('sqlite-shaped rows land correctly in real Postgres', () => {
+  let pgDb;
+  let svc;
+
+  beforeAll(async () => {
+    pgDb = knexLib({ client: 'pg', connection: PG_URL });
+    await pgDb.raw('DROP TABLE IF EXISTS xengine_events, xengine_settings CASCADE');
+    await pgDb.schema.createTable('xengine_events', (t) => {
+      t.increments('id');
+      t.string('slug');
+      t.boolean('is_active').defaultTo(true);
+      t.boolean('allow_downloads').defaultTo(true);
+      t.timestamp('created_at');
+      t.timestamp('expires_at');
+    });
+    await pgDb.schema.createTable('xengine_settings', (t) => {
+      t.increments('id');
+      t.string('setting_key').notNullable().unique();
+      t.jsonb('setting_value');
+    });
+
+    jest.resetModules();
+    jest.doMock('../../knexfile', () => ({ client: 'pg' }));
+    jest.doMock('../../src/database/db', () => ({ db: pgDb }));
+    svc = require('../../src/services/picpeakImportService');
+  });
+
+  afterAll(async () => {
+    jest.dontMock('../../src/database/db');
+    jest.dontMock('../../knexfile');
+    if (pgDb) {
+      await pgDb.raw('DROP TABLE IF EXISTS xengine_events, xengine_settings CASCADE');
+      await pgDb.destroy();
+    }
+  });
+
+  it('typedColumnsFor classifies boolean and timestamp columns via columnInfo()', async () => {
+    const types = await svc.typedColumnsFor(pgDb, 'xengine_events');
+    expect(types.booleans.sort()).toEqual(['allow_downloads', 'is_active']);
+    expect(types.timestamps.sort()).toEqual(['created_at', 'expires_at']);
+  });
+
+  it('inserts a sqlite archive row (0/1 booleans, epoch dates, json text) with correct stored values', async () => {
+    // Exactly what a sqlite-created .picpeak carries: integers for booleans,
+    // epoch numbers for #485-shape timestamps (ms here, seconds covered by the
+    // epochToIso unit), a "YYYY-MM-DD HH:MM:SS" string for clean ones, and
+    // json columns as TEXT (the crossEngine path skips serialiseJsonColumns —
+    // the text is already what pg wants).
+    const epoch = 1723400000000;
+    const eventRows = [
+      { id: 1, slug: 'wedding', is_active: 1, allow_downloads: 0, created_at: epoch, expires_at: '2026-09-01 12:00:00' },
+    ];
+    const settingRows = [{ id: 1, setting_key: 'brand', setting_value: '{"name":"PicPeak","dark":true}' }];
+
+    await pgDb.transaction(async (trx) => {
+      const evTypes = await svc.typedColumnsFor(trx, 'xengine_events');
+      await trx.batchInsert('xengine_events', svc.coerceForTargetEngine(eventRows, evTypes), 100);
+      const stTypes = await svc.typedColumnsFor(trx, 'xengine_settings');
+      await trx.batchInsert('xengine_settings', svc.coerceForTargetEngine(settingRows, stTypes), 100);
+    });
+
+    const ev = await pgDb('xengine_events').where({ id: 1 }).first();
+    expect(ev.is_active).toBe(true);          // 1 → true, not backwards (#1028 class)
+    expect(ev.allow_downloads).toBe(false);   // 0 → false
+    expect(new Date(ev.created_at).getTime()).toBe(epoch);
+    expect(new Date(ev.expires_at).toISOString().slice(0, 10)).toBe('2026-09-01');
+
+    const st = await pgDb('xengine_settings').where({ id: 1 }).first();
+    // jsonb parsed back by the driver — value intact, no double encoding.
+    expect(st.setting_value).toEqual({ name: 'PicPeak', dark: true });
+  });
+
+  it('id sequence works after explicit-id insert + resync (next natural insert)', async () => {
+    await svc.resyncSequences(['xengine_events']);
+    const [next] = await pgDb('xengine_events')
+      .insert({ slug: 'fresh', is_active: true })
+      .returning('id');
+    expect(Number(next.id || next)).toBe(2);
+  });
+});

backend/__tests__/integration/picpeakExport.test.js

@@ -0,0 +1,94 @@
+'use strict';
+
+// Validates the engine-neutral .picpeak export: it must produce a real zip with
+// a manifest + per-table NDJSON, exclude knex bookkeeping, and honour the photo
+// toggle. Uses the shared CRM DB harness (temp SQLite) — no docker needed.
+process.env.JWT_SECRET = process.env.JWT_SECRET || 'test-secret-at-least-32-characters-long!!';
+
+const fs = require('fs');
+const path = require('path');
+const StreamZip = require('node-stream-zip');
+const { bootCrmDb } = require('./helpers/crmDb');
+
+let db;
+let cleanup;
+let tmpDir;
+let createPicpeak;
+
+// bootCrmDb MUST run before requiring the service (which transitively requires
+// db.js) so the export reads this test's DB, not the default path.
+beforeAll(async () => {
+  ({ db, cleanup, tmpDir } = await bootCrmDb());
+  process.env.STORAGE_PATH = tmpDir; // isolate file collection to the temp dir
+  ({ createPicpeak } = require('../../src/services/picpeakExportService'));
+}, 120000);
+
+afterAll(async () => {
+  await cleanup();
+});
+
+async function readZip(filePath) {
+  const zip = new StreamZip.async({ file: filePath });
+  const entries = Object.keys(await zip.entries());
+  const manifest = JSON.parse((await zip.entryData('manifest.json')).toString('utf8'));
+  await zip.close();
+  return { entries, manifest };
+}
+
+describe('picpeak export (.picpeak logical export)', () => {
+  it('produces a .picpeak with a manifest and per-table NDJSON', async () => {
+    const { filePath, manifest } = await createPicpeak({ includePhotos: false });
+    try {
+      expect(filePath.endsWith('.picpeak')).toBe(true);
+      expect(fs.existsSync(filePath)).toBe(true);
+
+      expect(manifest.format).toBe(1);
+      expect(manifest.kind).toBe('picpeak-backup');
+      expect(manifest.database.engine).toBe('sqlite');
+      expect(manifest.options.includePhotos).toBe(false);
+      expect(manifest.contains_secrets).toBe(true);
+      // Migrations seed real tables (e.g. app_settings) — expect several.
+      expect(Object.keys(manifest.tables).length).toBeGreaterThan(0);
+      expect(Object.keys(manifest.tables)).toContain('app_settings');
+
+      const { entries, manifest: zipped } = await readZip(filePath);
+      expect(entries).toContain('manifest.json');
+      expect(entries.some((n) => n.startsWith('data/') && n.endsWith('.ndjson'))).toBe(true);
+      expect(entries).toContain('data/app_settings.ndjson');
+      // Manifest inside the zip matches the returned one.
+      expect(zipped.tables).toEqual(manifest.tables);
+    } finally {
+      fs.rmSync(path.dirname(filePath), { recursive: true, force: true });
+    }
+  });
+
+  it('never exports knex bookkeeping tables', async () => {
+    const { filePath, manifest } = await createPicpeak({ includePhotos: false });
+    try {
+      const names = Object.keys(manifest.tables);
+      expect(names).not.toContain('knex_migrations');
+      expect(names).not.toContain('knex_migrations_lock');
+    } finally {
+      fs.rmSync(path.dirname(filePath), { recursive: true, force: true });
+    }
+  });
+
+  it('row counts in the manifest match the NDJSON line counts', async () => {
+    // Insert a couple of settings so at least one table is non-empty.
+    await db('app_settings')
+      .insert({ setting_key: 'picpeak_export_test_a', setting_value: JSON.stringify('1'), setting_type: 'string' })
+      .onConflict('setting_key').merge();
+
+    const { filePath, manifest } = await createPicpeak({ includePhotos: false });
+    try {
+      const zip = new StreamZip.async({ file: filePath });
+      const buf = await zip.entryData('data/app_settings.ndjson');
+      await zip.close();
+      const lines = buf.toString('utf8').split('\n').filter((l) => l.trim().length > 0);
+      expect(lines.length).toBe(manifest.tables.app_settings.rowCount);
+      expect(manifest.tables.app_settings.rowCount).toBeGreaterThan(0);
+    } finally {
+      fs.rmSync(path.dirname(filePath), { recursive: true, force: true });
+    }
+  });
+});

backend/__tests__/integration/picpeakRoundtrip.test.js

@@ -0,0 +1,180 @@
+'use strict';
+
+// Full .picpeak roundtrip on a temp SQLite DB:
+//   1. seed a "backup" instance (admin A + a marker setting)
+//   2. export → .picpeak
+//   3. simulate a reinstall: wipe, create a DIFFERENT current admin B, mutate data
+//   4. import the backup with currentAdminId = B
+//   5. assert the backup data is restored AND the current account (B) survives,
+//      while the backup's admin (A) is also present (different email → added).
+process.env.JWT_SECRET = process.env.JWT_SECRET || 'test-secret-at-least-32-characters-long!!';
+
+const fs = require('fs');
+const path = require('path');
+const { bootCrmDb } = require('./helpers/crmDb');
+
+let db;
+let cleanup;
+let tmpDir;
+let createPicpeak;
+let importFromPicpeak;
+let validateManifest;
+let superAdminRoleId;
+
+beforeAll(async () => {
+  ({ db, cleanup, tmpDir } = await bootCrmDb());
+  process.env.STORAGE_PATH = tmpDir;
+  ({ createPicpeak } = require('../../src/services/picpeakExportService'));
+  ({ importFromPicpeak, validateManifest } = require('../../src/services/picpeakImportService'));
+  const role = await db('roles').where({ name: 'super_admin' }).first();
+  superAdminRoleId = role.id;
+}, 120000);
+
+afterAll(async () => {
+  await cleanup();
+});
+
+const adminRow = (email, hash) => ({
+  username: email,
+  email,
+  password_hash: hash,
+  role_id: superAdminRoleId,
+  is_active: true,
+  must_change_password: false,
+  created_at: new Date(),
+  updated_at: new Date(),
+});
+
+async function setMarker(value) {
+  await db('app_settings')
+    .insert({ setting_key: 'roundtrip_marker', setting_value: JSON.stringify(value), setting_type: 'string' })
+    .onConflict('setting_key').merge();
+}
+async function getMarker() {
+  const row = await db('app_settings').where({ setting_key: 'roundtrip_marker' }).first();
+  return row ? JSON.parse(row.setting_value) : null;
+}
+
+describe('.picpeak roundtrip (export → import)', () => {
+  it('restores backup data and preserves the current account', async () => {
+    // 1. Seed the "source" instance.
+    await db('admin_users').del();
+    await db('admin_users').insert(adminRow('backup-admin@old.example', 'HASH_A'));
+    await setMarker('from_backup');
+
+    // 2. Export.
+    const { filePath } = await createPicpeak({ includePhotos: false });
+
+    try {
+      // 3. Simulate a reinstall: fresh current admin B, mutated data.
+      await db('admin_users').del();
+      const [bId] = await db('admin_users').insert(adminRow('current-admin@new.example', 'HASH_B')).returning('id');
+      const currentAdminId = typeof bId === 'object' ? bId.id : bId;
+      await setMarker('mutated_after_backup');
+
+      // 4. Import, preserving the current admin.
+      const result = await importFromPicpeak({ filePath: undefined, picpeakPath: filePath, currentAdminId });
+      expect(result.restored).toBe(true);
+      expect(result.tables).toBeGreaterThan(0);
+
+      // 5a. Backup data restored (marker reverted to the backup value).
+      expect(await getMarker()).toBe('from_backup');
+
+      // 5b. The backup's admin is present (different email → added).
+      const a = await db('admin_users').whereRaw('lower(email) = lower(?)', ['backup-admin@old.example']).first();
+      expect(a).toBeTruthy();
+      expect(a.password_hash).toBe('HASH_A');
+
+      // 5c. The current account SURVIVES the override, with its own credentials.
+      const b = await db('admin_users').whereRaw('lower(email) = lower(?)', ['current-admin@new.example']).first();
+      expect(b).toBeTruthy();
+      expect(b.password_hash).toBe('HASH_B');
+    } finally {
+      fs.rmSync(path.dirname(filePath), { recursive: true, force: true });
+    }
+  });
+
+  it('overwrites a backup admin that collides with the current account email', async () => {
+    // Source has an admin at the SAME email the current operator will use.
+    await db('admin_users').del();
+    await db('admin_users').insert(adminRow('shared@example.com', 'OLD_HASH'));
+    await setMarker('collision_case');
+    const { filePath } = await createPicpeak({ includePhotos: false });
+
+    try {
+      // Reinstall: current admin uses the same email but a NEW password.
+      await db('admin_users').del();
+      const [id] = await db('admin_users').insert(adminRow('shared@example.com', 'NEW_HASH')).returning('id');
+      const currentAdminId = typeof id === 'object' ? id.id : id;
+
+      await importFromPicpeak({ picpeakPath: filePath, currentAdminId });
+
+      // Exactly one admin at that email, and it keeps the CURRENT password.
+      const rows = await db('admin_users').whereRaw('lower(email) = lower(?)', ['shared@example.com']);
+      expect(rows).toHaveLength(1);
+      expect(rows[0].password_hash).toBe('NEW_HASH');
+    } finally {
+      fs.rmSync(path.dirname(filePath), { recursive: true, force: true });
+    }
+  });
+
+  it('restores files/ and reports filesRestored', async () => {
+    // A business-doc that lives in storage → travels in the backup.
+    const docDir = path.join(tmpDir, 'business-docs');
+    const marker = path.join(docDir, 'roundtrip-doc.txt');
+    fs.mkdirSync(docDir, { recursive: true });
+    fs.writeFileSync(marker, 'hello');
+    await db('admin_users').del();
+    const [id] = await db('admin_users').insert(adminRow('files@example.com', 'H')).returning('id');
+    const currentAdminId = typeof id === 'object' ? id.id : id;
+
+    const { filePath } = await createPicpeak({ includePhotos: false });
+    try {
+      fs.rmSync(marker); // delete on disk so the restore must bring it back
+      const result = await importFromPicpeak({ picpeakPath: filePath, currentAdminId });
+      expect(result.filesRestored).toBeGreaterThanOrEqual(1);
+      expect(fs.existsSync(marker)).toBe(true);
+      expect(fs.readFileSync(marker, 'utf8')).toBe('hello');
+    } finally {
+      fs.rmSync(path.dirname(filePath), { recursive: true, force: true });
+      fs.rmSync(docDir, { recursive: true, force: true });
+    }
+  });
+});
+
+describe('.picpeak manifest validation', () => {
+  it('rejects a database-engine mismatch', async () => {
+    // Harness runs on SQLite, so a pg manifest must be refused.
+    const blockers = await validateManifest({
+      kind: 'picpeak-backup', format: 1, database: { engine: 'pg' }, tables: {},
+    });
+    expect(blockers.some((b) => /engine/i.test(b))).toBe(true);
+  });
+
+  it('rejects a backup from a newer schema (forward-only)', async () => {
+    // validateManifest reads knex_migrations for the target's latest migration;
+    // the harness has none, so create it with an older migration than the backup.
+    await db.schema.createTable('knex_migrations', (t) => {
+      t.increments('id');
+      t.string('name');
+      t.integer('batch');
+      t.timestamp('migration_time');
+    });
+    try {
+      await db('knex_migrations').insert({ name: '100_baseline', batch: 1 });
+      const blockers = await validateManifest({
+        kind: 'picpeak-backup', format: 1,
+        database: { engine: 'sqlite', latest_migration: '999_from_the_future' },
+        tables: {},
+      });
+      expect(blockers.some((b) => /newer/i.test(b))).toBe(true);
+    } finally {
+      await db.schema.dropTableIfExists('knex_migrations');
+    }
+  });
+
+  it('rejects a file that is not a PicPeak backup', async () => {
+    const blockers = await validateManifest({ some: 'random-json' });
+    expect(blockers.length).toBeGreaterThan(0);
+  });
+});

backend/__tests__/integration/publishPasswordPolicy.test.js

@@ -0,0 +1,114 @@
+/**
+ * Publishing must not be a way around the configured gallery password policy.
+ *
+ * `POST /:id/publish` (#627) re-hashes `password_hash` from a plaintext the
+ * admin re-types in the publish dialog, and validated it with nothing but
+ * express-validator's `isLength({ min: 6 })`. So the configured complexity —
+ * moderate by default, meaning 8 characters plus upper, lower and a digit —
+ * governed event creation and password reset, while this door accepted
+ * `aaaaaa` and made it the live gallery password.
+ *
+ * Not an escalation: it needs admin auth plus events.edit, and such an admin
+ * could already set a weak password elsewhere. It is a policy gap — the admin
+ * UI advertises a complexity level this write path did not enforce.
+ */
+
+const path = require('path');
+const fs = require('fs');
+const os = require('os');
+
+process.env.NODE_ENV = 'test';
+process.env.TEST_DATABASE_PATH = path.join(
+  fs.mkdtempSync(path.join(os.tmpdir(), 'picpeak-pubpolicy-')), 'db.sqlite',
+);
+process.env.JWT_SECRET = process.env.JWT_SECRET || 'publish-policy-test-secret';
+process.env.STORAGE_PATH = fs.mkdtempSync(path.join(os.tmpdir(), 'picpeak-pubpolicy-storage-'));
+
+const request = require('supertest');
+const {
+  bootCrmDb, seedMinimal, assignAdminRole, mintAdminToken, buildRouteApp,
+} = require('./helpers/crmDb');
+
+describe('publish enforces the gallery password policy', () => {
+  let db; let cleanup; let app; let token;
+
+  beforeAll(async () => {
+    ({ db, cleanup } = await bootCrmDb());
+    const { adminId } = await seedMinimal(db);
+    await assignAdminRole(db, adminId, 'admin');
+    token = mintAdminToken(adminId);
+    app = buildRouteApp('/admin/events', require('../../src/routes/adminEvents'));
+  }, 120000);
+
+  afterAll(async () => { if (cleanup) await cleanup(); });
+
+  async function seedDraft(slug) {
+    const [row] = await db('events').insert({
+      slug,
+      event_type: 'wedding',
+      event_name: `Event ${slug}`,
+      event_date: '2026-09-01',
+      host_email: 'client@example.com',
+      admin_email: 'admin@example.com',
+      password_hash: 'original-hash',
+      require_password: 1,
+      share_link: `/gallery/${slug}/share`,
+      share_token: `${slug}-token`,
+      expires_at: new Date(Date.now() + 7 * 24 * 3600 * 1000).toISOString(),
+      is_active: 1,
+      is_archived: 0,
+      is_draft: 1,
+      created_at: new Date().toISOString(),
+    }).returning('id');
+    return typeof row === 'object' ? row.id : row;
+  }
+
+  it('refuses a password that misses the configured complexity', async () => {
+    const id = await seedDraft('weak-publish');
+
+    const res = await request(app)
+      .post(`/admin/events/${id}/publish`)
+      .set('Authorization', `Bearer ${token}`)
+      .send({ password: 'aaaaaa' });
+
+    expect(res.status).toBe(400);
+    expect(res.body.error).toMatch(/security requirements/i);
+
+    // Rejected BEFORE the write, not after — the gallery must be untouched,
+    // and still a draft.
+    const after = await db('events').where({ id }).first();
+    expect(after.password_hash).toBe('original-hash');
+    expect(after.is_draft === 1 || after.is_draft === true).toBe(true);
+  });
+
+  it('still accepts a password that meets it', async () => {
+    const id = await seedDraft('strong-publish');
+
+    const res = await request(app)
+      .post(`/admin/events/${id}/publish`)
+      .set('Authorization', `Bearer ${token}`)
+      .send({ password: 'Sup3r-Secret' });
+
+    expect(res.status).toBe(200);
+
+    const bcrypt = require('bcrypt');
+    const after = await db('events').where({ id }).first();
+    expect(after.password_hash).not.toBe('original-hash');
+    expect(await bcrypt.compare('Sup3r-Secret', after.password_hash)).toBe(true);
+  });
+
+  it('leaves a publish without a password alone', async () => {
+    // The legacy sentinel path: no password in the body means no rehash, so
+    // the policy has nothing to check and must not block the publish.
+    const id = await seedDraft('no-password-publish');
+
+    const res = await request(app)
+      .post(`/admin/events/${id}/publish`)
+      .set('Authorization', `Bearer ${token}`)
+      .send({});
+
+    expect(res.status).toBe(200);
+    const after = await db('events').where({ id }).first();
+    expect(after.password_hash).toBe('original-hash');
+  });
+});

backend/__tests__/integration/regenerateThumbnailsScript.test.js

@@ -0,0 +1,260 @@
+/**
+ * scripts/regenerate-thumbnails.js against external photos (#1148).
+ *
+ * The same defect #1129 fixed in the admin route, still standing in the CLI
+ * fallback: the script resolved every source as
+ * `storage/events/active/<photo.path>` and fs.access'd it. External and
+ * reference rows do not live there — their originals sit under
+ * `events.external_path` — so every one failed the check and was counted as an
+ * error. On an install where all photos are external the script did nothing at
+ * all, while reporting one error per photo.
+ *
+ * Driven against a REAL file on a REAL external mount with the real
+ * imageProcessor, not a mock: the whole point is that the source resolves off
+ * the mount, and a mocked ensureThumbnail would assert nothing about that.
+ *
+ * Responsive tiers (#1095/#1109) do not exist on this branch, so the tier
+ * backfill in the main twin has nothing to port. Everything else does.
+ */
+
+const fs = require('fs');
+const path = require('path');
+const os = require('os');
+const sharp = require('sharp');
+const { execFile } = require('child_process');
+
+describe('regenerate-thumbnails script (#1148)', () => {
+  let tmpDir; let db; let cleanup; let regenerateThumbnails;
+  let eventId; let externalPhotoId; let videoPhotoId; let watcherVideoId; let repairPhotoId;
+  let vanishingPhotoId;
+  let externalRoot;
+
+  beforeAll(async () => {
+    tmpDir = await fs.promises.mkdtemp(path.join(os.tmpdir(), 'picpeak-regen-script-'));
+    process.env.NODE_ENV = 'test';
+    process.env.TEST_DATABASE_PATH = path.join(tmpDir, 'data', 'test.db');
+    process.env.STORAGE_PATH = path.join(tmpDir, 'storage');
+    // External sources are sandboxed under EXTERNAL_MEDIA_ROOT. Rows carry a
+    // path relative to that root (#1163), so the 'wedding/' prefix on each
+    // external_relpath below is the event folder, not decoration.
+    process.env.EXTERNAL_MEDIA_ROOT = path.join(tmpDir, 'media');
+    externalRoot = path.join(process.env.EXTERNAL_MEDIA_ROOT, 'wedding');
+
+    await fs.promises.mkdir(path.dirname(process.env.TEST_DATABASE_PATH), { recursive: true });
+    await fs.promises.mkdir(process.env.STORAGE_PATH, { recursive: true });
+    await fs.promises.mkdir(externalRoot, { recursive: true });
+
+    jest.resetModules();
+    ({ db, cleanup } = await require('./helpers/crmDb').bootCrmDb());
+
+    // A real image on the external mount — never under events/active.
+    await sharp({
+      create: { width: 1200, height: 800, channels: 3, background: { r: 10, g: 90, b: 160 } },
+    }).jpeg().toFile(path.join(externalRoot, 'shot.jpg'));
+
+    const [ev] = await db('events').insert({
+      slug: 'regen-script-event',
+      event_type: 'wedding',
+      event_name: 'Regen Script',
+      event_date: '2026-08-01',
+      host_email: 'h@example.com',
+      admin_email: 'a@example.com',
+      password_hash: 'x',
+      share_link: '/gallery/regen-script-event/share',
+      expires_at: new Date(Date.now() + 7 * 24 * 3600 * 1000).toISOString(),
+      source_mode: 'reference',
+      external_path: 'wedding',
+      created_at: new Date().toISOString(),
+    }).returning('id');
+    eventId = typeof ev === 'object' ? ev.id : ev;
+
+    const [p] = await db('photos').insert({
+      event_id: eventId,
+      filename: 'shot.jpg',
+      // `path` is what the old script joined onto events/active. Left
+      // populated on purpose: the fix must ignore it for an external row.
+      path: 'regen-script-event/shot.jpg',
+      type: 'individual',
+      source_origin: 'external',
+      external_relpath: 'wedding/shot.jpg',
+      uploaded_at: new Date().toISOString(),
+    }).returning('id');
+    externalPhotoId = typeof p === 'object' ? p.id : p;
+
+    const [v] = await db('photos').insert({
+      event_id: eventId,
+      filename: 'clip.mp4',
+      path: 'regen-script-event/clip.mp4',
+      type: 'individual',
+      media_type: 'video',
+      mime_type: 'video/mp4',
+      source_origin: 'external',
+      external_relpath: 'wedding/clip.mp4',
+      uploaded_at: new Date().toISOString(),
+    }).returning('id');
+    videoPhotoId = typeof v === 'object' ? v.id : v;
+
+    // How fileWatcher.processNewPhoto actually writes a video: `type` and
+    // `mime_type` set, media_type left to its 'image' default. A media_type-only
+    // filter lets this through and hands the container to Sharp.
+    //
+    // The file has to EXIST, otherwise the row fails resolution and looks
+    // skipped for the wrong reason — the bug is Sharp being handed a video, not
+    // a missing source. Real MP4 header bytes, no image in sight.
+    await fs.promises.writeFile(
+      path.join(externalRoot, 'watched.mp4'),
+      Buffer.from('00000018667479706d70343200000000', 'hex')
+    );
+    const [wv] = await db('photos').insert({
+      event_id: eventId,
+      filename: 'watched.mp4',
+      path: 'regen-script-event/watched.mp4',
+      type: 'video',
+      mime_type: 'video/mp4',
+      source_origin: 'external',
+      external_relpath: 'wedding/watched.mp4',
+      uploaded_at: new Date().toISOString(),
+    }).returning('id');
+    watcherVideoId = typeof wv === 'object' ? wv.id : wv;
+    expect((await db('photos').where('id', watcherVideoId).first()).media_type).not.toBe('video');
+
+    // A photo whose thumbnail_path points at something that is no longer there.
+    await sharp({
+      create: { width: 900, height: 600, channels: 3, background: { r: 200, g: 40, b: 40 } },
+    }).jpeg().toFile(path.join(externalRoot, 'repair.jpg'));
+    const [rp] = await db('photos').insert({
+      event_id: eventId,
+      filename: 'repair.jpg',
+      path: 'regen-script-event/repair.jpg',
+      type: 'individual',
+      thumbnail_path: 'thumbnails/thumb_ext_missing_repair.jpg',
+      source_origin: 'external',
+      external_relpath: 'wedding/repair.jpg',
+      uploaded_at: new Date().toISOString(),
+    }).returning('id');
+    repairPhotoId = typeof rp === 'object' ? rp.id : rp;
+
+    // A photo whose source is not on the mount at all — an unavailable mount,
+    // which is the failure an operator most needs to hear about.
+    const [vp] = await db('photos').insert({
+      event_id: eventId,
+      filename: 'missing.jpg',
+      path: 'regen-script-event/missing.jpg',
+      type: 'individual',
+      source_origin: 'external',
+      external_relpath: 'missing.jpg',
+      uploaded_at: new Date().toISOString(),
+    }).returning('id');
+    vanishingPhotoId = typeof vp === 'object' ? vp.id : vp;
+
+    ({ regenerateThumbnails } = require('../../scripts/regenerate-thumbnails'));
+  }, 180000);
+
+  afterAll(async () => {
+    if (cleanup) await cleanup();
+    await fs.promises.rm(tmpDir, { recursive: true, force: true }).catch(() => {});
+  });
+
+  it('builds a thumbnail for an external photo instead of erroring on events/active', async () => {
+    // The location the old script computed and fs.access'd. Nothing is there,
+    // which is the whole defect — it is not where an external original lives.
+    // (The old script cannot be driven from a test directly: it had no export
+    // and ran on require, calling process.exit. Making it importable is part
+    // of this fix.)
+    const legacyPath = path.join(process.env.STORAGE_PATH, 'events/active', 'regen-script-event/shot.jpg');
+    expect(fs.existsSync(legacyPath)).toBe(false);
+
+    const result = await regenerateThumbnails(eventId);
+
+    // The old script reported an error for this photo and wrote nothing.
+    // The unresolvable row fails; the external photo and the repair row build.
+    expect(result.errorCount).toBe(1);
+    expect(result.successCount).toBe(2);
+
+    const row = await db('photos').where('id', externalPhotoId).first();
+    expect(row.thumbnail_path).toBeTruthy();
+    const onDisk = path.join(process.env.STORAGE_PATH, row.thumbnail_path);
+    expect(fs.existsSync(onDisk)).toBe(true);
+
+    // Named per-photo so two events referencing one NAS basename cannot
+    // clobber each other — the property ensureThumbnail owns and the reason
+    // the script must not build this name itself.
+    expect(path.basename(row.thumbnail_path)).toContain(`ext${externalPhotoId}_`);
+  });
+
+  it('leaves videos alone', async () => {
+    // A video thumbnail is a poster frame from videoProcessor; handing the
+    // container to Sharp produced one error per video row.
+    const row = await db('photos').where('id', videoPhotoId).first();
+    expect(row.thumbnail_path).toBeFalsy();
+  });
+
+  it('leaves a watcher-imported video alone, which carries no media_type', async () => {
+    // fileWatcher writes type + mime_type and lets media_type default to
+    // 'image', so filtering on media_type alone still fed these to Sharp. The
+    // signal is errorCount: the images are already done by now, so the only
+    // NEW thing that could fail this run is a video reaching Sharp. One error
+    // is the deliberately unresolvable row; two would be the video.
+    const result = await regenerateThumbnails(eventId);
+
+    expect(result.errorCount).toBe(1);
+    const row = await db('photos').where('id', watcherVideoId).first();
+    expect(row.thumbnail_path).toBeFalsy();
+  });
+
+  it('is idempotent — a second run skips instead of rebuilding', async () => {
+    const before = await db('photos').where('id', externalPhotoId).first();
+    const result = await regenerateThumbnails(eventId);
+
+    expect(result.errorCount).toBe(1);
+    expect(result.successCount).toBe(0);
+    expect(result.skipCount).toBe(2);
+
+    const after = await db('photos').where('id', externalPhotoId).first();
+    expect(after.thumbnail_path).toBe(before.thumbnail_path);
+  });
+
+  it('counts a repaired thumbnail as generated, not skipped', async () => {
+    // Both images are valid at this point. Destroy ONE thumbnail object while
+    // leaving thumbnail_path pointing at it — the corrupt/missing case.
+    const row = await db('photos').where('id', repairPhotoId).first();
+    const onDisk = path.join(process.env.STORAGE_PATH, row.thumbnail_path);
+    await fs.promises.rm(onDisk);
+
+    const result = await regenerateThumbnails(eventId);
+
+    // On local and external storage the rebuilt key is identical, so inferring
+    // "skipped" from an unchanged path reports this repair as already valid —
+    // the one number an operator running this is actually reading.
+    expect(result.successCount).toBe(1);
+    expect(result.skipCount).toBe(1);
+    expect(result.errorCount).toBe(1);
+    expect(fs.existsSync(onDisk)).toBe(true);
+  });
+
+  /** Run the CLI the way cron does, and hand back its exit status. */
+  const runCli = (args = []) => new Promise((resolve) => {
+    execFile(
+      process.execPath,
+      [path.join(__dirname, '..', '..', 'scripts', 'regenerate-thumbnails.js'), ...args],
+      { env: { ...process.env }, cwd: path.join(__dirname, '..', '..') },
+      (error, stdout, stderr) => resolve({ code: error?.code ?? 0, stdout, stderr })
+    );
+  });
+
+  it('exits nonzero when a photo could not be built', async () => {
+    // Exit status is the only thing a cron job reads, and `missing.jpg` has no
+    // source on the mount.
+    const failed = await runCli([String(eventId)]);
+    expect(failed.code).toBe(1);
+    expect(failed.stderr).toContain('completed with failures');
+  }, 120000);
+
+  it('exits zero when every photo resolves', async () => {
+    // Drop the unresolvable row: a clean run must not cry wolf at automation.
+    await db('photos').where('id', vanishingPhotoId).del();
+    const ok = await runCli([String(eventId)]);
+    expect(ok.code).toBe(0);
+    expect(ok.stdout).toContain('Script completed successfully');
+  }, 120000);
+});

backend/__tests__/integration/resetAdminMfaCli.test.js

@@ -0,0 +1,82 @@
+/**
+ * CLI test for scripts/reset-admin-mfa.js — break-glass MFA reset (#738).
+ *
+ * Boots a temp-SQLite DB, seeds an admin with MFA fully enabled, then runs
+ * the script in a child process (--email <addr> --yes) pointed at the same
+ * DB file, and asserts the four MFA columns are zeroed. The script runs in
+ * its own process with its own knex connection; the parent connection is
+ * idle during the spawn so the SQLite write lock isn't contended.
+ */
+
+const path = require('path');
+const { execFileSync } = require('child_process');
+
+const { bootCrmDb } = require('./helpers/crmDb');
+
+jest.setTimeout(120000);
+
+let db;
+let cleanup;
+
+beforeAll(async () => {
+  ({ db, cleanup } = await bootCrmDb());
+}, 120000);
+
+afterAll(async () => {
+  if (cleanup) await cleanup();
+});
+
+const SCRIPT = path.resolve(__dirname, '..', '..', 'scripts', 'reset-admin-mfa.js');
+
+async function seedEnrolledAdmin(email) {
+  const inserted = await db('admin_users').insert({
+    username: email.split('@')[0],
+    email,
+    password_hash: 'x',
+    is_active: true,
+    two_factor_enabled: true,
+    two_factor_secret: 'iv.tag.ct',
+    two_factor_recovery_codes: JSON.stringify(['$2b$10$fakehashfakehashfakehashfa']),
+    two_factor_enrolled_at: new Date(),
+    created_at: new Date(),
+  }).returning('id');
+  return inserted[0]?.id ?? inserted[0];
+}
+
+it('zeroes the four MFA columns for the targeted admin', async () => {
+  const email = 'reset-me@example.com';
+  const id = await seedEnrolledAdmin(email);
+
+  execFileSync('node', [SCRIPT, '--email', email, '--yes'], {
+    env: {
+      ...process.env,
+      NODE_ENV: 'test',
+      TEST_DATABASE_PATH: process.env.TEST_DATABASE_PATH,
+    },
+    stdio: 'pipe',
+  });
+
+  const row = await db('admin_users').where({ id }).first();
+  expect(Number(row.two_factor_enabled)).toBe(0);
+  expect(row.two_factor_secret).toBeNull();
+  expect(row.two_factor_recovery_codes).toBeNull();
+  expect(row.two_factor_enrolled_at).toBeNull();
+});
+
+it('leaves a different admin untouched', async () => {
+  const targetEmail = 'target@example.com';
+  const bystanderEmail = 'bystander@example.com';
+  const targetId = await seedEnrolledAdmin(targetEmail);
+  const bystanderId = await seedEnrolledAdmin(bystanderEmail);
+
+  execFileSync('node', [SCRIPT, '--email', targetEmail, '--yes'], {
+    env: { ...process.env, NODE_ENV: 'test', TEST_DATABASE_PATH: process.env.TEST_DATABASE_PATH },
+    stdio: 'pipe',
+  });
+
+  const target = await db('admin_users').where({ id: targetId }).first();
+  const bystander = await db('admin_users').where({ id: bystanderId }).first();
+  expect(Number(target.two_factor_enabled)).toBe(0);
+  expect(Number(bystander.two_factor_enabled)).toBe(1);
+  expect(bystander.two_factor_secret).toBe('iv.tag.ct');
+});

backend/__tests__/integration/restoreService.pgBranch.test.js

@@ -0,0 +1,261 @@
+/**
+ * Pins the fix for the PR #596 review blocker.
+ *
+ * **The bug**
+ *
+ *   `preservedMeta` was declared with `let` INSIDE the PostgreSQL
+ *   `else` branch of `performDatabaseRestore`, then read AFTER the
+ *   `else` block closed at the shared replay site (~L1030). On every
+ *   real PG restore:
+ *
+ *     ReferenceError: preservedMeta is not defined
+ *
+ *   would fire — psql had already completed the data restore, but
+ *   the operator-meta replay never ran, the trigger file was left
+ *   in place by `_installFromBackupBoot.js` because the restore
+ *   "failed", and `combined.log` got a loud FAILED line even though
+ *   the data was back. Caught on PR #596 review by the maintainer.
+ *
+ * **Why CI missed it**
+ *
+ *   The integration tests around `performFullRestore` only exercise
+ *   the SQLite branch via `this.dbType === 'sqlite'`. The PG branch
+ *   (~L827-984) requires a real PG connection + real `psql` binary,
+ *   neither of which are in the test environment. So the scope leak
+ *   sat untested until the maintainer ran a real DR cycle.
+ *
+ * **What this test does**
+ *
+ *   Reads the source of `restoreService.js` and asserts the scope
+ *   contract: the `preservedMeta` declaration sits ABOVE the
+ *   SQLite/PG branch split, so the replay block at the bottom of the
+ *   try{} can read it on either branch.
+ *
+ *   Source-inspection is uglier than a runtime test but it has two
+ *   advantages here: (a) it doesn't require a real PG cluster + psql
+ *   binary in CI, (b) it pins the EXACT contract — "the declaration
+ *   must be visible to the replay block" — which is the property
+ *   that broke, more directly than a runtime test would.
+ *
+ *   The follow-up "real-PG integration test in CI" (separate task)
+ *   would replace this with an end-to-end exercise, at which point
+ *   this can be deleted.
+ */
+
+const fs = require('fs');
+const path = require('path');
+
+describe('restoreService — PG branch scope contract (PR #596 review)', () => {
+  let src;
+  let lines;
+
+  beforeAll(() => {
+    src = fs.readFileSync(
+      path.join(__dirname, '..', '..', 'src', 'services', 'restoreService.js'),
+      'utf8',
+    );
+    lines = src.split(/\r?\n/);
+  });
+
+  /** Return the 1-based line number of the FIRST line matching `re`. */
+  function findFirst(re) {
+    const idx = lines.findIndex((l) => re.test(l));
+    return idx >= 0 ? idx + 1 : -1;
+  }
+
+  /** Return the 1-based line number of the LAST line matching `re`. */
+  function findLast(re) {
+    let last = -1;
+    lines.forEach((l, i) => { if (re.test(l)) last = i + 1; });
+    return last;
+  }
+
+  it('preservedMetaSnapshot lives on `this` and is initialised in the constructor', () => {
+    // PR #596 round 3 moved the snapshot from a block-scoped local to
+    // an instance variable so the replay can happen in `restore()`
+    // AFTER post-restore verification — preventing the replay row
+    // from inflating the row-count check.
+    //
+    // Contract:
+    //   1. The constructor initialises `this.preservedMetaSnapshot = []`
+    //   2. The `restore()` entry point resets it per call (no leak
+    //      across consecutive runs in the singleton service instance)
+    //   3. `performDatabaseRestore` assigns to `this.preservedMetaSnapshot`
+    //      inside the PG branch (must run before DROP)
+    //   4. The replay reads `this.preservedMetaSnapshot` — NOT a bare
+    //      `preservedMeta` local — so a future refactor can't
+    //      accidentally drop the snapshot half on the floor again.
+    const constructorInit = lines.some((l) =>
+      /this\.preservedMetaSnapshot\s*=\s*\[\s*\]/.test(l)
+    );
+    expect(constructorInit).toBe(true);
+
+    const assignmentSites = lines.filter((l) =>
+      /this\.preservedMetaSnapshot\s*=\s*(\[\s*\]|await\s+db)/.test(l)
+    );
+    // Constructor init + restore() per-run reset + the PG-branch
+    // assignment from db query. Three writes.
+    expect(assignmentSites.length).toBeGreaterThanOrEqual(3);
+
+    // No stray bare `preservedMeta` local-scoped declaration in
+    // performDatabaseRestore — would indicate someone re-introduced
+    // the round-1 footgun.
+    const dangerousLocalDecl = lines.filter((l) =>
+      /^\s*(let|const)\s+preservedMeta\s*=/.test(l)
+    );
+    expect(dangerousLocalDecl).toEqual([]);
+  });
+
+  it('every .count() result is coerced to Number before comparison', () => {
+    // PR #596 review caught a second PG-only landmine: pg-driver
+    // returns COUNT(*) as a string ("16" not 16) to preserve bigint
+    // precision. The original code compared `result.count !==
+    // expected.rowCount` and every match flagged as a mismatch on PG.
+    //
+    // The fix coerces with `Number(...)` at every comparison +
+    // interpolation site. This test catches a future regression where
+    // a refactor uses `.count` directly in a `===` / `!==` / `>` /
+    // `<` comparison without coercing.
+    //
+    // Heuristic: find every `.count` access in the file and make sure
+    // the line either:
+    //   (a) wraps it in `Number(...)`, or
+    //   (b) is purely an interpolation that already coerced upstream
+    //       (e.g. `validation.warnings.push(`... ${eventCountN} ...`)`
+    //       where eventCountN is the coerced local), or
+    //   (c) is the docstring/comment line (filtered separately).
+    //
+    // We approximate this by listing every `.count` reference site
+    // and asserting that lines doing comparisons (`===`/`!==`/`>`/
+    // `<`/`>=`/`<=`) on a raw `.count` access without `Number(...)`
+    // around it are zero.
+    const dangerousLines = lines
+      .map((l, i) => ({ line: i + 1, text: l }))
+      // Filter to lines that compare a .count result
+      .filter(({ text }) => {
+        // Skip comments
+        if (/^\s*(\/\/|\*)/.test(text)) return false;
+        // Detect a `.count` (followed by `)` for `?.count` or by space/operator)
+        // being directly compared via ===/!==/>/<.
+        // Match the BAD pattern: `<something>.count <op> <something>`
+        // where <op> is === / !== / > / < / >= / <=
+        const bareCountInComparison = /\w+\??\.count\s*(?:!==|===|>=?|<=?)\s+/;
+        // ALLOW if the .count is preceded by `Number(` in the same line
+        const wrappedInNumber = /Number\(\s*\w+\??\.count/;
+        return bareCountInComparison.test(text) && !wrappedInNumber.test(text);
+      });
+
+    expect(dangerousLines).toEqual([]);
+  });
+
+  it('the completed-restore update sets was_successful=true', () => {
+    // Without this, every successful restore ends up with
+    // status='completed', was_successful=false — the dashboard's
+    // "last successful restore" widget then filters out the row +
+    // any future audit query gating on was_successful misses it.
+    // Caught locally + maintainer PR #596 review.
+    //
+    // Contract: the update payload that writes status='completed' on
+    // the SUCCESS branch ALSO includes was_successful: true. We pin
+    // it by source inspection so any future refactor of the success
+    // payload keeps both fields together.
+    // The success-branch update lives AFTER performPostRestoreVerification.
+    // There's also a `status: 'completed'` in the dry-run / early-return
+    // path (failure handling has its own block too) — we want the
+    // SUCCESS-branch one specifically.
+    const verifyLine = findFirst(/performPostRestoreVerification\s*\(/);
+    expect(verifyLine).toBeGreaterThan(0);
+
+    const completedStatusLineIdx = lines
+      .map((l, i) => ({ line: i + 1, text: l }))
+      .find(({ line, text }) =>
+        line > verifyLine && /status:\s*['"]completed['"]/.test(text)
+      );
+    expect(completedStatusLineIdx).toBeDefined();
+
+    // Look in the next ~10 lines for was_successful: true. The actual
+    // payload is small (no nested objects between status and the
+    // closing })), so a fixed-window search is reliable.
+    const window = lines.slice(
+      completedStatusLineIdx.line - 1,
+      completedStatusLineIdx.line + 10,
+    ).join('\n');
+    expect(window).toMatch(/was_successful:\s*true/);
+  });
+
+  it('the safe migration runner is invoked after the replay in restore()', () => {
+    // Contract from PR #596 round 4: backups taken on older picpeak
+    // versions must restore COMPLETELY on a newer image — even if new
+    // migrations have been added since the backup was taken. The
+    // restore() flow shells out to the safe migration runner AFTER the
+    // operator-meta replay so the schema catches up to the running
+    // code WITHIN the restore boundary (not on the next container
+    // restart). Invoked as `node migrations/run-migrations-safe.js` —
+    // the runtime image ships no npm, so the former `npm run
+    // migrate:safe` would ENOENT into the non-fatal catch.
+    //
+    // Contract:
+    //   1. A run-migrations-safe shell-out exists somewhere in restoreService
+    //   2. It sits AFTER the replay drain — verification → replay →
+    //      migrations is the documented order
+    //   3. It does NOT sit inside performDatabaseRestore (must run
+    //      against the reinit'd pool from the parent restore())
+    const migrateLine = findFirst(/run-migrations-safe\.js/);
+    expect(migrateLine).toBeGreaterThan(0);
+
+    const replayLine = findLast(/this\.preservedMetaSnapshot\.length\s*>\s*0/);
+    expect(replayLine).toBeGreaterThan(0);
+    expect(migrateLine).toBeGreaterThan(replayLine);
+
+    // Must NOT live inside performDatabaseRestore (same scope as the
+    // replay check above).
+    const dbRestoreStart = findFirst(/async\s+performDatabaseRestore\s*\(/);
+    let dbRestoreEnd = -1;
+    for (let i = dbRestoreStart; i < lines.length; i++) {
+      if (/^  \}\s*$/.test(lines[i])) {
+        dbRestoreEnd = i + 1;
+        break;
+      }
+    }
+    expect(migrateLine < dbRestoreStart || migrateLine > dbRestoreEnd).toBe(true);
+  });
+
+  it('the replay site lives in restore() AFTER performPostRestoreVerification', () => {
+    // PR #596 round 3 moved the replay out of performDatabaseRestore
+    // and into the parent restore() method, sequenced AFTER the
+    // post-restore verification. Otherwise the replay's upserted row
+    // count was being flagged as a verification mismatch (e.g.
+    // "expected 190, got 191" because the fresh-install seeded
+    // `restore_allow_force_auto_upgraded` that wasn't in the backup).
+    //
+    // Contract: the line that drains `this.preservedMetaSnapshot`
+    // must come AFTER `performPostRestoreVerification` AND must NOT
+    // sit inside `performDatabaseRestore`.
+    const verificationLine = findFirst(/performPostRestoreVerification\s*\(/);
+    expect(verificationLine).toBeGreaterThan(0);
+
+    const replayLine = findLast(/this\.preservedMetaSnapshot\.length\s*>\s*0/);
+    expect(replayLine).toBeGreaterThan(0);
+    expect(replayLine).toBeGreaterThan(verificationLine);
+
+    // `performDatabaseRestore` must not contain the replay drain.
+    // Find the function bounds + assert no drain line falls inside.
+    const dbRestoreStart = findFirst(/async\s+performDatabaseRestore\s*\(/);
+    expect(dbRestoreStart).toBeGreaterThan(0);
+
+    // Find the closing brace of performDatabaseRestore. Lazy heuristic:
+    // the first `^  \}\s*$` (two-space indent + }) after the function
+    // start. Brittle to indent changes but unambiguous in this codebase.
+    let dbRestoreEnd = -1;
+    for (let i = dbRestoreStart; i < lines.length; i++) {
+      if (/^  \}\s*$/.test(lines[i])) {
+        dbRestoreEnd = i + 1;
+        break;
+      }
+    }
+    expect(dbRestoreEnd).toBeGreaterThan(dbRestoreStart);
+
+    // The replay drain line must be OUTSIDE [dbRestoreStart, dbRestoreEnd].
+    expect(replayLine < dbRestoreStart || replayLine > dbRestoreEnd).toBe(true);
+  });
+});

backend/__tests__/integration/setupService.test.js

@@ -0,0 +1,196 @@
+'use strict';
+
+// First-run bootstrap service. bootCrmDb() must run BEFORE requiring the service
+// so setupService shares this test's db instance (see crmDb.js note).
+process.env.JWT_SECRET = process.env.JWT_SECRET || 'test-secret-at-least-32-characters-long!!';
+
+const fs = require('fs');
+const path = require('path');
+const request = require('supertest');
+const { bootCrmDb, buildRouteApp } = require('./helpers/crmDb');
+
+let db;
+let cleanup;
+let tmpDir;
+let setupService;
+let getAppSetting;
+let upsertAppSetting;
+let app;
+
+const VALID_PW = 'Str0ng-Passw0rd!';
+
+// bootCrmDb MUST run before any require of db.js (directly or transitively via a
+// service/util), or db.js binds to the default path instead of the temp one.
+beforeAll(async () => {
+  ({ db, cleanup, tmpDir } = await bootCrmDb());
+  process.env.DATA_DIR = tmpDir; // isolate the SETUP_TOKEN file to the temp dir
+  setupService = require('../../src/services/setupService');
+  ({ getAppSetting, upsertAppSetting } = require('../../src/utils/appSettings'));
+  app = buildRouteApp('/api/setup', require('../../src/routes/setup'));
+}, 120000);
+
+afterAll(async () => {
+  await cleanup();
+});
+
+beforeEach(async () => {
+  await db('admin_users').del();
+  await db('app_settings').where({ setting_key: 'setup_token' }).del();
+});
+
+describe('setupService (first-run bootstrap)', () => {
+  it('reports needsAdmin while no admin exists', async () => {
+    expect(await setupService.getSetupStatus()).toEqual({ needsAdmin: true, complete: false });
+  });
+
+  it('generates and persists a one-time token while no admin exists', async () => {
+    const token = await setupService.ensureSetupToken();
+    expect(token).toEqual(expect.any(String));
+    expect(token.length).toBeGreaterThan(20);
+    expect(await getAppSetting('setup_token')).toBe(token);
+    // Idempotent — a second call returns the same token, not a fresh one.
+    expect(await setupService.ensureSetupToken()).toBe(token);
+  });
+
+  it('stores the token as valid JSON so the Postgres jsonb column accepts it', async () => {
+    // Regression guard for the SQLite-only miss: a bare token string is rejected
+    // by Postgres jsonb ("invalid input syntax for type json"). The raw column
+    // value must be JSON-parseable and round-trip back to the token.
+    const token = await setupService.ensureSetupToken();
+    const row = await db('app_settings').where({ setting_key: 'setup_token' }).first();
+    expect(() => JSON.parse(row.setting_value)).not.toThrow();
+    expect(JSON.parse(row.setting_value)).toBe(token);
+  });
+
+  it('rejects a wrong token', async () => {
+    await setupService.ensureSetupToken();
+    await expect(
+      setupService.createInitialAdmin({ token: 'nope', email: 'a@b.co', password: VALID_PW })
+    ).rejects.toMatchObject({ statusCode: 400 });
+    expect(await setupService.getSetupStatus()).toEqual({ needsAdmin: true, complete: false });
+  });
+
+  it('rejects a weak password', async () => {
+    const token = await setupService.ensureSetupToken();
+    await expect(
+      setupService.createInitialAdmin({ token, email: 'a@b.co', password: 'weak' })
+    ).rejects.toMatchObject({ statusCode: 400 });
+  });
+
+  it('creates the first admin as super_admin, issues a token, and burns the setup token', async () => {
+    const token = await setupService.ensureSetupToken();
+    const result = await setupService.createInitialAdmin({
+      token, email: 'Owner@Example.com', password: VALID_PW, ip: '203.0.113.7',
+    });
+
+    expect(result.user.email).toBe('owner@example.com'); // normalised
+    expect(result.user.role.name).toBe('super_admin');
+    expect(result.token).toEqual(expect.any(String));
+
+    const row = await db('admin_users').first();
+    const role = await db('roles').where({ name: 'super_admin' }).first();
+    expect(row.role_id).toBe(role.id);
+    expect(row.password_hash).not.toBe(VALID_PW); // hashed
+
+    // One-time: token burned, status now complete.
+    expect(await getAppSetting('setup_token')).toBeFalsy();
+    expect(await setupService.getSetupStatus()).toEqual({ needsAdmin: false, complete: true });
+  });
+
+  it('writes the SETUP_TOKEN file while pending and removes it once setup completes', async () => {
+    const tokenFile = path.join(tmpDir, 'SETUP_TOKEN');
+    const token = await setupService.ensureSetupToken();
+    expect(fs.readFileSync(tokenFile, 'utf8').trim()).toBe(token);
+    await setupService.createInitialAdmin({ token, email: 'owner@example.com', password: VALID_PW });
+    expect(fs.existsSync(tokenFile)).toBe(false); // burned in DB + file removed
+  });
+
+  it('refuses to create a second admin (setup already complete)', async () => {
+    const token = await setupService.ensureSetupToken();
+    await setupService.createInitialAdmin({ token, email: 'first@example.com', password: VALID_PW });
+    await expect(
+      setupService.createInitialAdmin({ token, email: 'second@example.com', password: VALID_PW })
+    ).rejects.toMatchObject({ statusCode: 409 });
+  });
+
+  it('serialises a double-submit — two concurrent valid-token calls create only one admin', async () => {
+    const token = await setupService.ensureSetupToken();
+    const results = await Promise.allSettled([
+      setupService.createInitialAdmin({ token, email: 'a@example.com', password: VALID_PW }),
+      setupService.createInitialAdmin({ token, email: 'b@example.com', password: VALID_PW }),
+    ]);
+    const fulfilled = results.filter((r) => r.status === 'fulfilled');
+    expect(fulfilled).toHaveLength(1); // the atomic token claim lets exactly one win
+    const count = await db('admin_users').count({ c: '*' }).first();
+    expect(Number(count.c)).toBe(1);
+  });
+
+  it('ensureSetupToken clears any stale token once an admin exists', async () => {
+    const token = await setupService.ensureSetupToken();
+    await setupService.createInitialAdmin({ token, email: 'first@example.com', password: VALID_PW });
+    // Simulate a stale token left in settings, then re-run the boot hook.
+    await upsertAppSetting('setup_token', JSON.stringify('stale'), 'string');
+    expect(await setupService.ensureSetupToken()).toBeNull();
+    expect(await getAppSetting('setup_token')).toBeFalsy();
+  });
+});
+
+describe('setup routes', () => {
+  it('GET /api/setup/status reports needsAdmin', async () => {
+    const res = await request(app).get('/api/setup/status');
+    expect(res.status).toBe(200);
+    expect(res.body).toEqual({ needsAdmin: true, complete: false });
+  });
+
+  it('POST /api/setup/verify-token accepts the right token without burning it (200)', async () => {
+    const token = await setupService.ensureSetupToken();
+    const res = await request(app).post('/api/setup/verify-token').send({ token });
+    expect(res.status).toBe(200);
+    expect(res.body).toEqual({ valid: true });
+    // Token is NOT consumed — it still works for the actual create.
+    expect(await getAppSetting('setup_token')).toBe(token);
+  });
+
+  it('POST /api/setup/verify-token rejects a wrong token (400, field token)', async () => {
+    await setupService.ensureSetupToken();
+    const res = await request(app).post('/api/setup/verify-token').send({ token: 'nope' });
+    expect(res.status).toBe(400);
+    expect(res.body.field).toBe('token');
+  });
+
+  it('POST /api/setup/verify-token is closed once an admin exists (409)', async () => {
+    const token = await setupService.ensureSetupToken();
+    await setupService.createInitialAdmin({ token, email: 'first@example.com', password: VALID_PW });
+    const res = await request(app).post('/api/setup/verify-token').send({ token });
+    expect(res.status).toBe(409);
+  });
+
+  it('POST /api/setup/admin rejects a wrong token (400)', async () => {
+    await setupService.ensureSetupToken();
+    const res = await request(app)
+      .post('/api/setup/admin')
+      .send({ token: 'nope', email: 'a@b.co', password: VALID_PW });
+    expect(res.status).toBe(400);
+    expect(await setupService.getSetupStatus()).toMatchObject({ needsAdmin: true });
+  });
+
+  it('POST /api/setup/admin creates the first admin + sets the auth cookie (201)', async () => {
+    const token = await setupService.ensureSetupToken();
+    const res = await request(app)
+      .post('/api/setup/admin')
+      .send({ token, email: 'owner@example.com', password: VALID_PW });
+    expect(res.status).toBe(201);
+    expect(res.body.user.role.name).toBe('super_admin');
+    expect((res.headers['set-cookie'] || []).join(';')).toMatch(/admin_token/);
+    expect(await setupService.getSetupStatus()).toEqual({ needsAdmin: false, complete: true });
+  });
+
+  it('POST /api/setup/admin is closed once an admin exists (409)', async () => {
+    const token = await setupService.ensureSetupToken();
+    await setupService.createInitialAdmin({ token, email: 'first@example.com', password: VALID_PW });
+    const res = await request(app)
+      .post('/api/setup/admin')
+      .send({ token, email: 'second@example.com', password: VALID_PW });
+    expect(res.status).toBe(409);
+  });
+});

backend/__tests__/integration/slideshowPhotoSource.test.js

@@ -0,0 +1,150 @@
+/**
+ * Slideshow photo source (#1015).
+ *
+ * The bug: with `lightbox_preview_enabled` off (the default), /photos emitted
+ * `preview_url: null`, so the slideshow's `preview_url || hero_url || url`
+ * chain fell through to `hero_url` — a 1920x1080 `fit: 'cover'` centre crop
+ * meant for gallery header banners. With the "Black Bars (No crop)" fit the
+ * show then letterboxed an already-cropped frame: portrait photos lost their
+ * top and bottom and the setting looked broken.
+ *
+ * The contract pinned here: `slideshow_url` points at the aspect-preserved
+ * preview tier and is emitted for image photos REGARDLESS of the lightbox
+ * toggle, so the slideshow never has a reason to reach for `hero_url`.
+ * `preview_url` itself must stay gated — the lightbox opt-in is unchanged.
+ */
+
+const request = require('supertest');
+const express = require('express');
+const cookieParser = require('cookie-parser');
+const jwt = require('jsonwebtoken');
+
+const { bootCrmDb, seedMinimal } = require('./helpers/crmDb');
+
+process.env.JWT_SECRET = process.env.JWT_SECRET || 'slideshow-src-test-secret';
+
+const SLUG = 'slideshow-source-event';
+
+describe('Slideshow photo source (#1015)', () => {
+  let db;
+  let cleanup;
+  let app;
+  let eventId;
+  let imagePhotoId;
+  let videoPhotoId;
+
+  const galleryToken = () => jwt.sign(
+    { eventId, eventSlug: SLUG, type: 'gallery' },
+    process.env.JWT_SECRET,
+    { expiresIn: '1h', issuer: 'picpeak-auth' }
+  );
+
+  const setLightboxPreview = async (on) => {
+    await db('app_settings').where({ setting_key: 'lightbox_preview_enabled' }).del();
+    await db('app_settings').insert({
+      setting_key: 'lightbox_preview_enabled',
+      setting_value: JSON.stringify(on),
+      setting_type: 'general',
+      updated_at: new Date().toISOString(),
+    });
+  };
+
+  const fetchPhotos = async () => {
+    const res = await request(app)
+      .get(`/api/gallery/${SLUG}/photos`)
+      .set('Authorization', `Bearer ${galleryToken()}`)
+      .expect(200);
+    return res.body.photos;
+  };
+
+  beforeAll(async () => {
+    ({ db, cleanup } = await bootCrmDb());
+    await seedMinimal(db);
+
+    const inserted = await db('events').insert({
+      slug: SLUG,
+      event_type: 'wedding',
+      event_name: 'Slideshow Source Test',
+      event_date: '2026-08-01',
+      host_email: 'host@example.com',
+      admin_email: 'admin@example.com',
+      password_hash: 'x',
+      share_link: `/gallery/${SLUG}/share`,
+      share_token: 'slideshow-source-share',
+      expires_at: new Date(Date.now() + 7 * 24 * 3600 * 1000).toISOString(),
+      is_active: 1,
+      is_archived: 0,
+      is_draft: 0,
+      created_at: new Date().toISOString(),
+    }).returning('id');
+    eventId = inserted[0]?.id ?? inserted[0];
+
+    const img = await db('photos').insert({
+      event_id: eventId,
+      filename: 'portrait.jpg',
+      path: 'events/slideshow-source/portrait.jpg',
+      type: 'individual',
+      mime_type: 'image/jpeg',
+      uploaded_at: new Date().toISOString(),
+    }).returning('id');
+    imagePhotoId = img[0]?.id ?? img[0];
+
+    const vid = await db('photos').insert({
+      event_id: eventId,
+      filename: 'clip.mp4',
+      path: 'events/slideshow-source/clip.mp4',
+      type: 'individual',
+      media_type: 'video',
+      mime_type: 'video/mp4',
+      uploaded_at: new Date().toISOString(),
+    }).returning('id');
+    videoPhotoId = vid[0]?.id ?? vid[0];
+
+    app = express();
+    app.use(express.json());
+    app.use(cookieParser());
+    app.use('/api/gallery', require('../../src/routes/gallery'));
+  }, 120000);
+
+  afterAll(async () => {
+    if (cleanup) await cleanup();
+  });
+
+  it('emits slideshow_url for image photos even when lightbox previews are OFF', async () => {
+    await setLightboxPreview(false);
+    const photos = await fetchPhotos();
+    const image = photos.find((p) => p.id === imagePhotoId);
+
+    expect(image.slideshow_url).toBe(`/api/gallery/${SLUG}/preview/${imagePhotoId}`);
+    // The regression: this is what used to be null, pushing the show to hero.
+    expect(image.preview_url).toBeNull();
+  });
+
+  it('leaves preview_url gated so the lightbox opt-in is unchanged', async () => {
+    await setLightboxPreview(true);
+    const photos = await fetchPhotos();
+    const image = photos.find((p) => p.id === imagePhotoId);
+
+    expect(image.preview_url).toBe(`/api/gallery/${SLUG}/preview/${imagePhotoId}`);
+    expect(image.slideshow_url).toBe(image.preview_url);
+  });
+
+  it('never points the slideshow at the cover-cropped hero tier', async () => {
+    await setLightboxPreview(false);
+    const photos = await fetchPhotos();
+    const image = photos.find((p) => p.id === imagePhotoId);
+
+    // hero_url still ships (the gallery header uses it) — it just must not be
+    // what the slideshow resolves to.
+    expect(image.hero_url).toBe(`/api/gallery/${SLUG}/hero/${imagePhotoId}`);
+    expect(image.slideshow_url).not.toBe(image.hero_url);
+  });
+
+  it('emits slideshow_url: null for videos, which have no preview tier', async () => {
+    await setLightboxPreview(false);
+    const photos = await fetchPhotos();
+    const video = photos.find((p) => p.id === videoPhotoId);
+
+    expect(video.slideshow_url).toBeNull();
+  });
+});

backend/__tests__/integration/storageBackend.test.js

@@ -0,0 +1,189 @@
+const path = require('path');
+const fs = require('fs');
+const fsp = require('fs').promises;
+const os = require('os');
+const crypto = require('crypto');
+const { Readable } = require('stream');
+const { S3Client, CreateBucketCommand, DeleteBucketCommand, ListObjectsV2Command, DeleteObjectsCommand } = require('@aws-sdk/client-s3');
+
+const LocalFsStorage = require('../../src/services/storage/LocalFsStorage');
+const S3StorageBackend = require('../../src/services/storage/S3StorageBackend');
+
+// MinIO defaults match docker-compose.dev.yml. Override via TEST_S3_* if needed.
+const TEST_S3 = {
+  endpoint: process.env.TEST_S3_ENDPOINT || 'http://localhost:7104',
+  accessKeyId: process.env.TEST_S3_ACCESS_KEY || 'minioadmin',
+  secretAccessKey: process.env.TEST_S3_SECRET_KEY || 'minioadmin',
+  region: 'us-east-1',
+};
+
+const skipS3 = process.env.SKIP_S3_TESTS === 'true';
+
+// Build the matrix of backends to test. Local always runs; S3 runs against MinIO
+// unless SKIP_S3_TESTS=true (CI default). The same suite runs against both so
+// every consumer can rely on identical semantics.
+function backendCases() {
+  const cases = [
+    {
+      name: 'LocalFsStorage',
+      async setup() {
+        const root = await fsp.mkdtemp(path.join(os.tmpdir(), 'picpeak-storage-'));
+        const storage = new LocalFsStorage({ root });
+        await storage.init();
+        return { storage, cleanup: () => fsp.rm(root, { recursive: true, force: true }) };
+      },
+    },
+  ];
+
+  if (!skipS3) {
+    cases.push({
+      name: 'S3StorageBackend (MinIO)',
+      async setup() {
+        const bucket = `picpeak-test-${Date.now()}-${crypto.randomBytes(2).toString('hex')}`;
+        const s3Client = new S3Client({
+          endpoint: TEST_S3.endpoint,
+          region: TEST_S3.region,
+          credentials: { accessKeyId: TEST_S3.accessKeyId, secretAccessKey: TEST_S3.secretAccessKey },
+          forcePathStyle: true,
+        });
+        await s3Client.send(new CreateBucketCommand({ Bucket: bucket }));
+        const storage = new S3StorageBackend({
+          bucket,
+          region: TEST_S3.region,
+          endpoint: TEST_S3.endpoint,
+          accessKeyId: TEST_S3.accessKeyId,
+          secretAccessKey: TEST_S3.secretAccessKey,
+          forcePathStyle: true,
+          sslEnabled: false,
+        });
+        await storage.init();
+        return {
+          storage,
+          async cleanup() {
+            // Empty bucket then delete it.
+            const list = await s3Client.send(new ListObjectsV2Command({ Bucket: bucket }));
+            if (list.Contents?.length) {
+              await s3Client.send(new DeleteObjectsCommand({
+                Bucket: bucket,
+                Delete: { Objects: list.Contents.map((o) => ({ Key: o.Key })) },
+              }));
+            }
+            await s3Client.send(new DeleteBucketCommand({ Bucket: bucket }));
+          },
+        };
+      },
+    });
+  }
+
+  return cases;
+}
+
+async function readToString(stream) {
+  const chunks = [];
+  for await (const chunk of stream) chunks.push(Buffer.isBuffer(chunk) ? chunk : Buffer.from(chunk));
+  return Buffer.concat(chunks).toString('utf-8');
+}
+
+describe.each(backendCases())('StorageBackend contract: $name', ({ setup }) => {
+  let storage;
+  let cleanup;
+
+  beforeAll(async () => {
+    ({ storage, cleanup } = await setup());
+  }, 30000);
+
+  afterAll(async () => {
+    if (cleanup) await cleanup();
+  });
+
+  test('put + get + exists + stat + delete round-trip with a buffer body', async () => {
+    const key = 'photos/event-a/IMG_0001.jpg';
+    const body = Buffer.from('hello picpeak');
+
+    await storage.put(key, body, { contentType: 'image/jpeg' });
+
+    expect(await storage.exists(key)).toBe(true);
+
+    const stat = await storage.stat(key);
+    expect(stat).not.toBeNull();
+    expect(stat.size).toBe(body.length);
+
+    const stream = await storage.get(key);
+    const text = await readToString(stream);
+    expect(text).toBe('hello picpeak');
+
+    await storage.delete(key);
+    expect(await storage.exists(key)).toBe(false);
+    expect(await storage.stat(key)).toBeNull();
+  });
+
+  test('put accepts a Readable stream body', async () => {
+    const key = 'photos/event-b/streamed.bin';
+    const body = Readable.from(Buffer.from('streamed payload'));
+
+    await storage.put(key, body);
+
+    const got = await readToString(await storage.get(key));
+    expect(got).toBe('streamed payload');
+  });
+
+  test('putFromFile + getToFile round-trip', async () => {
+    const tmpIn = path.join(os.tmpdir(), `in-${Date.now()}.txt`);
+    const tmpOut = path.join(os.tmpdir(), `out-${Date.now()}.txt`);
+    await fsp.writeFile(tmpIn, 'file payload');
+
+    const key = 'thumbnails/thumb_x.jpg';
+    await storage.putFromFile(key, tmpIn, { contentType: 'image/jpeg' });
+
+    await storage.getToFile(key, tmpOut);
+    const text = await fsp.readFile(tmpOut, 'utf-8');
+    expect(text).toBe('file payload');
+
+    await fsp.unlink(tmpIn).catch(() => {});
+    await fsp.unlink(tmpOut).catch(() => {});
+  });
+
+  test('list returns entries under a prefix with size + key', async () => {
+    await storage.put('events/active/a/photo1.jpg', Buffer.from('a1'));
+    await storage.put('events/active/a/photo2.jpg', Buffer.from('a22'));
+    await storage.put('events/active/b/photo3.jpg', Buffer.from('b333'));
+
+    const entries = await storage.list('events/active/a');
+    const keys = entries.map((e) => e.key).sort();
+    expect(keys).toEqual(['events/active/a/photo1.jpg', 'events/active/a/photo2.jpg']);
+    const sizes = Object.fromEntries(entries.map((e) => [e.key, e.size]));
+    expect(sizes['events/active/a/photo1.jpg']).toBe(2);
+    expect(sizes['events/active/a/photo2.jpg']).toBe(3);
+  });
+
+  test('rename moves an object from src to dst (atomic on local; copy+delete on s3)', async () => {
+    await storage.put('uploads/temp.jpg', Buffer.from('rename-me'));
+    await storage.rename('uploads/temp.jpg', 'uploads/final.jpg');
+
+    expect(await storage.exists('uploads/temp.jpg')).toBe(false);
+    expect(await storage.exists('uploads/final.jpg')).toBe(true);
+    const text = await readToString(await storage.get('uploads/final.jpg'));
+    expect(text).toBe('rename-me');
+  });
+
+  test('copy duplicates an object without removing the source', async () => {
+    await storage.put('events/source.jpg', Buffer.from('src'));
+    await storage.copy('events/source.jpg', 'events/copied.jpg');
+
+    expect(await storage.exists('events/source.jpg')).toBe(true);
+    expect(await storage.exists('events/copied.jpg')).toBe(true);
+  });
+
+  test('delete on a missing key is a no-op (does not throw)', async () => {
+    await expect(storage.delete('does/not/exist.jpg')).resolves.toBeUndefined();
+  });
+
+  test('stat on a missing key returns null', async () => {
+    expect(await storage.stat('still/not/here.jpg')).toBeNull();
+  });
+
+  test('rejects path traversal attempts', async () => {
+    await expect(storage.put('../escape.txt', Buffer.from('x'))).rejects.toThrow(/traversal/i);
+    await expect(storage.get('../escape.txt')).rejects.toThrow(/traversal/i);
+  });
+});

backend/__tests__/integration/uncategorizedFilter.test.js

@@ -0,0 +1,107 @@
+/**
+ * The admin photo list's category filter, and the value it answers to (#1211).
+ *
+ * The frontend used to send `category_id=0` for "Uncategorized". This route
+ * skips `'0'` outright — the guard reads `category_id !== '0'` — so no
+ * condition was applied and the whole event came back. Four lines below that
+ * guard sits the branch that does the work, keyed on the literal
+ * `uncategorized`, which nothing was sending.
+ *
+ * Reported in #1209 by someone trying to isolate a few thousand uncategorised
+ * imports. The frontend half is fixed in PhotoFilters; this pins the backend
+ * half of the same contract, because the failure mode was the two ends
+ * disagreeing about a string and neither one being wrong on its own.
+ */
+
+const request = require('supertest');
+const express = require('express');
+
+const { bootCrmDb, seedMinimal } = require('./helpers/crmDb');
+
+describe('admin photo list — uncategorized filter (#1211)', () => {
+  let db; let cleanup; let app;
+  let eventId; let categoryId;
+  let uncategorisedIds; let categorisedId;
+
+  const list = async (query = '') => {
+    const res = await request(app).get(`/api/admin/events/${eventId}/photos${query}`);
+    expect(res.status).toBe(200);
+    const photos = Array.isArray(res.body) ? res.body : res.body.photos;
+    return (photos || []).map((p) => p.id).sort((a, b) => a - b);
+  };
+
+  beforeAll(async () => {
+    jest.resetModules();
+    jest.doMock('../../src/middleware/auth', () => ({
+      adminAuth: (req, _res, next) => { req.admin = { id: 1, username: 'tester' }; next(); },
+    }));
+    jest.doMock('../../src/middleware/permissions', () => ({
+      requirePermission: () => (_req, _res, next) => next(),
+    }));
+    jest.doMock('../../src/middleware/ownership', () => ({
+      requireEventOwnership: (_req, _res, next) => next(),
+    }));
+    jest.doMock('../../src/utils/logger', () => ({
+      debug: jest.fn(), info: jest.fn(), warn: jest.fn(), error: jest.fn(),
+    }));
+
+    ({ db, cleanup } = await bootCrmDb());
+    await seedMinimal(db);
+
+    const [ev] = await db('events').insert({
+      slug: 'uncat-filter', event_type: 'wedding', event_name: 'Uncat Filter',
+      event_date: '2026-08-01', host_email: 'h@example.com', admin_email: 'a@example.com',
+      password_hash: 'x', share_link: '/gallery/uncat-filter/share',
+      expires_at: new Date(Date.now() + 7 * 24 * 3600 * 1000).toISOString(),
+      is_active: 1, is_archived: 0, is_draft: 0, created_at: new Date().toISOString(),
+    }).returning('id');
+    eventId = typeof ev === 'object' ? ev.id : ev;
+
+    const [cat] = await db('photo_categories')
+      .insert({ name: 'Ceremony', slug: 'ceremony', event_id: eventId })
+      .returning('id');
+    categoryId = typeof cat === 'object' ? cat.id : cat;
+
+    const insertPhoto = async (filename, category) => {
+      const [p] = await db('photos').insert({
+        event_id: eventId, filename, path: `events/uncat/${filename}`,
+        type: 'individual', category_id: category,
+        uploaded_at: new Date().toISOString(),
+      }).returning('id');
+      return typeof p === 'object' ? p.id : p;
+    };
+
+    // Two with no category — the shape a plugin upload leaves behind — and one
+    // filed properly, so a filter that does nothing is visibly different from
+    // a filter that works.
+    uncategorisedIds = [await insertPhoto('a.jpg', null), await insertPhoto('b.jpg', null)];
+    categorisedId = await insertPhoto('c.jpg', categoryId);
+    uncategorisedIds.sort((a, b) => a - b);
+
+    app = express();
+    app.use(express.json());
+    app.use('/api/admin/events', require('../../src/routes/adminPhotos'));
+  }, 180000);
+
+  afterAll(async () => { if (cleanup) await cleanup(); });
+
+  it('returns only the photos with no category', async () => {
+    expect(await list('?category_id=uncategorized')).toEqual(uncategorisedIds);
+  });
+
+  it('returns everything when no category filter is given', async () => {
+    expect(await list()).toEqual([...uncategorisedIds, categorisedId].sort((a, b) => a - b));
+  });
+
+  it('still filters by a real category id', async () => {
+    expect(await list(`?category_id=${categoryId}`)).toEqual([categorisedId]);
+  });
+
+  it('treats 0 as no filter at all', async () => {
+    // Pinning the behaviour that made the bug silent rather than loud: '0' is
+    // not "uncategorized" and never was, it simply falls through the guard. A
+    // future change that made 0 mean uncategorized here would be fine too —
+    // but it must be a decision, not an accident, and this test forces it.
+    expect(await list('?category_id=0')).toEqual([...uncategorisedIds, categorisedId].sort((a, b) => a - b));
+  });
+});