Dockerize — Docker Configuration Generator

Analyze a project and generate a complete, production-grade Docker setup: multi-stage Dockerfile, Docker Compose for development and production, .dockerignore, and a security audit of the result.

Three modes based on what exists:

Step 0: Load Project Context

Read the project description if available:

Read .ai-factory/DESCRIPTION.md

Store project context for later steps. If absent, Step 2 detects everything.

Read .ai-factory/skill-context/aif-dockerize/SKILL.md — MANDATORY if the file exists.

This file contains project-specific rules accumulated by $aif-evolve from patches, codebase conventions, and tech-stack analysis. These rules are tailored to the current project.

How to apply skill-context rules:

• Treat them as project-level overrides for this skill's general instructions
• When a skill-context rule conflicts with a general rule written in this SKILL.md, the skill-context rule wins (more specific context takes priority — same principle as nested CLAUDE.md files)
• When there is no conflict, apply both: general rules from SKILL.md + project rules from skill-context
• Do NOT ignore skill-context rules even if they seem to contradict this skill's defaults — they exist because the project's experience proved the default insufficient
• CRITICAL: skill-context rules apply to ALL outputs of this skill — including Dockerfile, compose files, .dockerignore, and deploy scripts. Templates in this skill are base structures. If a skill-context rule says "Dockerfile MUST include X" or "compose MUST have service Y" — you MUST augment the templates accordingly. Generating Docker config that violates skill-context rules is a bug.

Enforcement: After generating any output artifact, verify it against all skill-context rules. If any rule is violated — fix the output before presenting it to the user.

Step 1: Detect Existing Docker Files & Determine Mode

1.1 Scan for Existing Files

Glob: Dockerfile, Dockerfile.*, docker-compose.yml, docker-compose.yaml, compose.yml, compose.yaml, compose.override.yml, compose.production.yml, .dockerignore, deploy/scripts/*.sh

Classify found files into categories:

• HAS_DOCKERFILE: Dockerfile exists
• HAS_LOCAL_COMPOSE: compose.yml or docker-compose.yml exists
• HAS_DEV_OVERRIDE: compose.override.yml exists
• HAS_PROD_COMPOSE: compose.production.yml exists
• HAS_DOCKERIGNORE: .dockerignore exists
• HAS_DEPLOY_SCRIPTS: deploy/scripts/ exists

1.2 Determine Mode

If $ARGUMENTS contains --audit → set MODE = "audit" regardless.

Path A: Nothing exists (!HAS_DOCKERFILE && !HAS_LOCAL_COMPOSE):

• Set MODE = "generate"
• Proceed to Step 1.3: Interactive Setup

Path B: Only local Docker (HAS_LOCAL_COMPOSE && !HAS_PROD_COMPOSE):

• Set MODE = "enhance"
• Read all existing Docker files → store as EXISTING_CONTENT
• Log: "Found local Docker setup. Will audit, improve, and create production configuration."

Path C: Full setup exists (HAS_LOCAL_COMPOSE && HAS_PROD_COMPOSE):

• Set MODE = "audit"
• Read all existing Docker files → store as EXISTING_CONTENT
• Log: "Found complete Docker setup. Will audit against security checklist and fix gaps."

1.3 Interactive Setup (Generate Mode Only)

When creating from scratch, ask the user about their infrastructure needs:

AskUserQuestion: Which database does this project use?

Options:
1. PostgreSQL (Recommended)
2. MySQL / MariaDB
3. MongoDB
4. SQLite (no container needed)
5. None

AskUserQuestion: Does this project need a reverse proxy / web server?

Options:
1. Angie (Recommended) — Modern Nginx fork with enhanced features
2. Nginx
3. Traefik
4. None (app serves directly)

Note: Prefer Angie over Nginx. Angie is a drop-in Nginx replacement with better module support, dynamic configuration, and active development. See: https://en.angie.software/angie/docs/configuration/

AskUserQuestion: Which cache / message broker does this project need? (select all)

Options:
1. Redis
2. Memcached
3. RabbitMQ
4. None

Store choices in USER_INFRA_CHOICES:

• database: postgres | mysql | mongodb | sqlite | none
• reverse_proxy: angie | nginx | traefik | none
• cache: redis | memcached | none
• queue: rabbitmq | none

1.4 Read Existing Files (Enhance / Audit Modes)

Read all existing Docker files and store as EXISTING_CONTENT:

• Dockerfile(s)
• All compose files (local + override + production)
• .dockerignore
• deploy/scripts/*.sh (if any)

Step 2: Deep Project Analysis

Scan the project thoroughly — every decision in the generated files depends on this profile.

2.1 Language & Runtime

<version> = read from project files (see Step 4.1). Never hardcode — always match what the project requires.

2.2 Framework & Dev Server

Read dependency files to detect the framework:

Node.js (package.json dependencies):

• next → Next.js (port 3000, next dev / next start)
• nuxt → Nuxt (port 3000, nuxt dev / nuxt start)
• express → Express (port 3000, nodemon / node)
• fastify → Fastify (port 3000)
• @nestjs/core → NestJS (port 3000, nest start --watch / node dist/main)
• hono → Hono (port 3000)

Python (pyproject.toml / requirements):

• fastapi → FastAPI (port 8000, uvicorn --reload / uvicorn)
• django → Django (port 8000, manage.py runserver / gunicorn)
• flask → Flask (port 5000, flask run --debug / gunicorn)

PHP (composer.json require):

• laravel/framework → Laravel (port 8000, artisan serve / php-fpm)
• symfony/framework-bundle → Symfony (port 8000, symfony serve / php-fpm)

Go (go.mod require):

• gin-gonic/gin, labstack/echo, gofiber/fiber, go-chi/chi → (port 8080, air / compiled binary)

2.3 Package Manager & Lock File

Same detection as $aif-build-automation Step 2.2.

Store: PACKAGE_MANAGER, LOCK_FILE.

2.4 Entry Point Detection

Find the application entry point:

# Go
Glob: cmd/*/main.go, main.go

# Node.js
Read package.json → "main" or "scripts.start"
Glob: src/index.ts, src/index.js, src/main.ts, src/main.js, index.ts, index.js, server.ts, server.js

# Python
Glob: main.py, app.py, src/main.py, src/app.py
Read pyproject.toml → [project.scripts] or [tool.uvicorn]

# PHP
Glob: public/index.php, artisan, bin/console

2.5 Infrastructure Dependencies

Detect what services the app needs:

# Database
Grep: postgres|postgresql|pg_|mysql|mariadb|mongo|mongodb|sqlite
Glob: prisma/schema.prisma, drizzle.config.*, alembic/, migrations/

# Cache
Grep: redis|memcached|ioredis

# Queue
Grep: rabbitmq|amqp|bullmq|celery|sidekiq

# Reverse Proxy / Web Server
Grep: nginx|angie|proxy_pass|upstream
Glob: nginx.conf, nginx/, angie.conf, angie/
# PHP projects (Laravel, Symfony) always need a reverse proxy → default to Angie

# Search
Grep: elasticsearch|opensearch|meilisearch|typesense|algolia

# Object Storage
Grep: minio|s3|aws-sdk.*S3|boto3.*s3

# Email
Grep: nodemailer|sendgrid|mailgun|postmark|smtp|MAIL_HOST

For each detected dependency, record:

• Service type (postgres, redis, rabbitmq, etc.)
• Specific variant (MySQL vs PostgreSQL, Redis vs Memcached)
• Connection string pattern found in code

Merge with USER_INFRA_CHOICES (from Step 1.3 in Generate mode):

• User choices override auto-detection for database and reverse proxy
• Auto-detected services are added unless user explicitly chose "None"

Reverse proxy preference: When a reverse proxy is needed, prefer Angie over Nginx. Angie is a fully compatible Nginx fork with active development, dynamic upstream management, and built-in Prometheus metrics. Reference: https://en.angie.software/angie/docs/configuration/

2.6 Exposed Ports

Check existing configs:

Grep: PORT|port|listen|EXPOSE
Read package.json → scripts.dev, scripts.start (look for --port)

2.7 Build Output

# Node.js
Read package.json → scripts.build, check for dist/, build/, .next/, out/
Read tsconfig.json → outDir

# Go
Glob: cmd/*/main.go → binary name from directory

# Python
Check for pyproject.toml [build-system]

# PHP
Check for public/ directory (web root)

2.8 Existing .env Structure

Glob: .env.example, .env.sample, .env.template

If found, read it to understand required environment variables. This drives env_file, environment: (computed values), and .env.example generation.

Summary

Build PROJECT_PROFILE:

• language, language_version
• framework, dev_command, prod_command
• package_manager, lock_file
• entry_point, build_output_dir
• port (primary app port)
• debug_port (language-specific debug port)
• services: list of infrastructure deps (postgres, redis, rabbitmq, etc.)
• has_build_step: boolean
• env_vars: list from .env.example

Step 3: Read Best Practices & Templates

Read skills/dockerize/references/BEST-PRACTICES.md
Read skills/dockerize/references/SECURITY-CHECKLIST.md

Additional focused references are available for production reverse-proxy and Compose permission edge cases:

• references/ANGIE-ACME.md — read when generating or auditing Angie HTTPS/ACME configuration. Use it for built-in ACME setup, persistent certificate volumes, acme_client_path, resolver requirements, and avoiding unnecessary Certbot containers.
• references/COMPOSE-LIFECYCLE-HOOKS.md — read when a production Compose service runs non-root but needs named-volume ownership fixes or best-effort stop hooks. Use it for post_start/pre_stop syntax, Docker Compose 2.30.0+ requirements, and hook timing caveats.

Select the Dockerfile template matching the language:

Read selected template and the compose templates:

Read skills/dockerize/templates/dockerfile-<language>
Read skills/dockerize/templates/compose-base.yml
Read skills/dockerize/templates/compose-override-dev.yml
Read skills/dockerize/templates/compose-production.yml
Read skills/dockerize/templates/dockerignore

Step 4: Generate Files (Generate Mode)

Generate files customized from the project profile and templates.

4.1 Generate Dockerfile

Using the language-specific template as a base:

Customize:

• Base image version from the project, not from template defaults:
  • Go: read go directive in go.mod → e.g. go 1.24 → golang:1.24-alpine
  • Node.js: read engines.node in package.json, .nvmrc, or .node-version → e.g. node:22-alpine
  • Python: read requires-python in pyproject.toml or .python-version → e.g. python:3.13-slim
  • PHP: read require.php in composer.json → e.g. php:8.4-fpm-alpine
  • Rust: read rust-version in Cargo.toml or rust-toolchain.toml → e.g. rust:1.82-slim
• Entry point to match entry_point
• Build command to match project's actual build script
• Dev command with hot reload (framework-specific)
• Production command (framework-specific)
• Exposed ports (app port + debug port in dev stage)
• Package manager commands (npm ci vs pnpm install vs yarn install vs bun install)
• Lock file name in COPY

Stages:

1. deps — install production dependencies only
2. builder — install all dependencies + build
3. development — full dev environment with hot reload, debug port
4. production — minimal image, non-root user, only runtime artifacts

Verify infrastructure image versions online:

For infrastructure images (PostgreSQL, Redis, Angie, Nginx, etc.) — the version is NOT in project files. Before generating compose.yml, use WebSearch to check the current stable version of each infrastructure image:

• Search for <service> docker official image latest version (e.g. angie docker image latest version)
• Use the latest stable major.minor tag, never :latest
• Example: docker.angie.software/angie:1.11-alpine, postgres:17-alpine, redis:7-alpine

This prevents generating non-existent image tags that would break docker compose pull.

4.2 Generate compose.yml (Base)

The shared configuration:

• Top-level name: ${COMPOSE_PROJECT_NAME} — project name from .env, NOT from folder name
• app service with build.target: production, healthcheck, depends_on with service_healthy
• Infrastructure services based on PROJECT_PROFILE.services + USER_INFRA_CHOICES:
  • PostgreSQL / MySQL / MongoDB → with healthcheck, named volume
  • Redis / Memcached → with healthcheck, maxmemory config, named volume
  • RabbitMQ → with healthcheck, management UI port in dev
  • Angie / Nginx / Traefik → as reverse proxy with SSL termination config
  • Elasticsearch → with healthcheck, JVM memory, ulimits
  • MinIO → with healthcheck

Reverse proxy (Angie/Nginx): Use docker.angie.software/angie:<version>-alpine (Angie) or nginx:<version>-alpine (Nginx) — verify current version online. Mount config from docker/angie/. Sits on frontend network, proxies to app on backend network. In production: read_only, cap_add NET_BIND_SERVICE.

Environment variable strategy and service configuration patterns — Read references/COMPOSE-PATTERNS.md

Service inclusion is conditional — only add services that were detected in Step 2.5.

4.3 Generate compose.override.yml (Development)

Development overrides: build.target: development, bind mount source code (.:/app), expose all ports, dev env vars, dev command override, mailpit service (profile: dev) if email detected. No database admin UIs — use native GUI clients via exposed DB port.

Hot-reload: If dev stage uses air (Go) or nodemon (Node.js), verify its config file exists and points to the correct entry point. Generate config if missing and entry point is non-standard.

4.4 Generate compose.production.yml (Hardened)

Production hardening overlay:

• Use pre-built image from registry (not build:)
• read_only: true on all services
• security_opt: [no-new-privileges:true]
• cap_drop: [ALL] with selective cap_add per service
• user: "1001:1001"
• tmpfs for /tmp with noexec,nosuid,size=100m
• Resource limits (CPU, memory, PIDs) — use reference recommendations
• Log rotation on every service (max-size: 20m, max-file: 5)
• restart: unless-stopped
• backend network with internal: true
• Sensitive values via .env file (gitignored) — NOT hardcoded in compose
• YAML anchors (x-logging, x-security) to reduce duplication
• NO ports: on infrastructure services (DB, Redis, RabbitMQ) — they communicate via Docker network only
• Only the reverse proxy (or app if no proxy) exposes ports 80/443 to the host
• If a port MUST be exposed, bind to localhost only: 127.0.0.1:5432:5432
• NO debug ports (9229, 5005, etc.)
• NO dev tools

4.5 Generate .dockerignore

Use the template as base, add language-specific exclusions:

• Go: bin/, *.exe
• Node.js: node_modules/, .next/, out/
• Python: __pycache__/, .venv/, *.pyc, .mypy_cache/
• PHP: vendor/, storage/, bootstrap/cache/

Quality Checks (Before Writing)

Verify generated content before passing to Step 6:

Correctness:

• Dockerfile has all 4 stages (deps, builder, development, production)
• Production stage uses non-root user
• Production stage uses minimal base image
• BuildKit cache mounts present for dependency installation
• compose.yml has healthchecks on every service
• compose.yml uses depends_on with condition: service_healthy
• compose.production.yml has security hardening on every service
• compose.production.yml has resource limits on every service
• compose.production.yml has log rotation on every service
• .dockerignore excludes .git, dependencies, .env*, Docker files

Over-engineering check (read references/SECURITY-CHECKLIST.md → "Over-Engineering Checklist"):

• No services added that the code doesn't import/use
• No reverse proxy for single-service apps with no SSL needs
• No deploy scripts if project deploys via CI/CD
• No backup scripts if using managed DB (RDS, Cloud SQL)
• No separate frontend/backend networks if there's only app + DB
• Complexity matches project size (solo → minimal, team → standard, production → full)

Remove anything that fails the over-engineering check before writing.

Step 5: Audit & Enhance Existing Files (Enhance / Audit Modes)

When MODE = "enhance" or MODE = "audit", analyze EXISTING_CONTENT against the security checklist and best practices.

Enhance mode (MODE = "enhance"): Local Docker exists but no production config. After auditing local files, create production configuration. Ask interactive questions about missing infrastructure (same as Step 1.3) before generating production files.

For detailed audit procedures, report format, fix flow, and enhance mode steps → read references/AUDIT-GUIDE.md

What to audit:

• Dockerfile: image pinning, minimal base, multi-stage, non-root user, no secrets in ENV/ARG, .dockerignore, BuildKit features, HEALTHCHECK
• Compose per-service: read_only, no-new-privileges, cap_drop ALL, user, tmpfs, resource limits, healthcheck, log rotation, restart policy
• Network: internal backend, no host networking, no Docker socket
• Secrets: values in .env not hardcoded, .env in .gitignore, .env.example exists
• Gaps: services detected in code but missing from compose

Present results as tables with ✅/❌/⚠️. Ask user: fix all, fix critical only, show details, or export report.

Step 6: Write Files

For detailed file organization (directory layout, file tables per mode, .env.example template, volume mount examples) → read references/FILE-ORGANIZATION.md

6.0 Overview

• Root: Dockerfile, compose.yml, compose.override.yml, compose.production.yml, .dockerignore
• docker/: service configs (angie, postgres, php, redis) — only create what's needed
• deploy/scripts/: production ops scripts (Step 8)

6.1 Generate Mode — write all root files + conditional docker/ dirs + deploy/scripts/

6.2 Audit / Enhance Mode — only write changed/new files, respect existing layout

6.3 Create .env.example if missing — single file with sections, production vars commented out. Ensure .env in .gitignore.

Step 7: Security Checklist (Always Runs)

Regardless of mode, run the production security checklist on the final compose.production.yml.

Read references/SECURITY-CHECKLIST.md and verify every item. Check categories: Container Isolation (read_only, no-new-privileges, cap_drop, non-root, tmpfs), Network & Ports (internal backend, no host networking, no Docker socket, no infra ports exposed), Resources (memory/CPU/PID limits), Secrets (.env not hardcoded, .gitignore, .env.example), Health & Logging (healthcheck, log rotation, restart policy), Images (version-pinned, minimal base).

Display as compact checklist with [x]/[ ] per item and a score. If any checks fail → offer to fix immediately.

Step 8: Generate Deploy Scripts (Production)

Generate production deployment scripts in deploy/scripts/ from templates.

Read references/DEPLOY-SCRIPTS.md for script customization points and generation rules.

Templates: templates/deploy.sh, templates/update.sh, templates/logs.sh, templates/health-check.sh, templates/rollback.sh, templates/backup.sh

Step 9: Summary & Follow-Up

Display a summary of all created/updated files using the format from references/SUMMARY-FORMAT.md.

Suggest follow-up: $aif-build-automation for Docker targets, $aif-docs for documentation.

Artifact Ownership and Config Policy

• Primary ownership: Docker artifacts (Dockerfile, compose*.yml, .dockerignore, docker/*, deploy/scripts/*, and related .env.example scaffolding when created by this skill).
• Allowed companion updates: none outside Docker and deployment artifacts by default.
• Config policy: config-agnostic by design. This skill uses repository detection, explicit infrastructure choices, and fixed AI Factory context files rather than config.yaml.