Scaffold Docker

Scaffold Docker & Health Checks

You are setting up production-grade containerization for a mission-critical TypeScript project. This skill generates a hardened Dockerfile, health check module, and Docker Compose configuration.

Available Components

Instructions

1. Determine components

If $ARGUMENTS is provided, parse it as a comma-separated list. Otherwise, ask the user which components to set up. Default: all.

2. Load context

• Read package.json for project name, main entry point, scripts
• Read tsconfig.json for outDir (default: dist/)
• Check if src/index.ts or src/main.ts exists (application entry point)
• Check for existing Dockerfile, docker-compose.yml, src/health/

Component: dockerfile — Dockerfile + .dockerignore

Ask the user which base image they prefer for the production stage:

• node:22-slim (recommended for most cases) — smaller than full image, includes shell for debugging
• gcr.io/distroless/nodejs22-debian12 — no shell, no package manager, minimal attack surface (best for production)

Generate Dockerfile:

# =============================================================================
# Build Stage
# =============================================================================
FROM node:22-slim AS builder

WORKDIR /app

# Copy dependency files first for layer caching
COPY package.json package-lock.json ./

# Use npm ci for reproducible installs (Rule 3.4)
RUN npm ci

# Copy source and build
COPY tsconfig.json ./
COPY src/ ./src/
RUN npm run build

# Prune dev dependencies
RUN npm ci --omit=dev

# =============================================================================
# Production Stage
# =============================================================================
FROM <chosen-base-image> AS production

# Security: run as non-root user
USER node

WORKDIR /app

# Copy only production artifacts
COPY --from=builder --chown=node:node /app/dist ./dist
COPY --from=builder --chown=node:node /app/node_modules ./node_modules
COPY --from=builder --chown=node:node /app/package.json ./

# Environment
ENV NODE_ENV=production

# Health check (adjust port and path as needed)
HEALTHCHECK --interval=30s --timeout=5s --start-period=10s --retries=3 \
  CMD ["node", "-e", "fetch('http://localhost:3000/healthz').then(r => process.exit(r.ok ? 0 : 1)).catch(() => process.exit(1))"]

EXPOSE 3000

# Use node directly (not npm start) for proper signal handling
CMD ["node", "dist/index.js"]

If the user chose distroless:

• Remove HEALTHCHECK (no shell available; rely on Kubernetes probes)
• CMD uses JSON array format only
• No USER directive needed (distroless defaults to non-root)

If the user chose slim:

• Add dumb-init for PID 1 signal handling:

  RUN apt-get update && apt-get install -y --no-install-recommends dumb-init && rm -rf /var/lib/apt/lists/*
  ENTRYPOINT ["dumb-init", "--"]
  

Generate .dockerignore:

# Source and tests (built artifacts are in dist/)
src/
tests/
**/*.test.ts
**/*.spec.ts

# Documentation
docs/
*.md
!README.md

# Development
.git/
.github/
.claude/
.vscode/
.idea/
coverage/
.stryker-tmp/
reports/

# Environment and secrets
.env
.env.*
*.pem
*.key

# Build tools
stryker.config.*
vitest.config.*
tsconfig.json
.eslintrc*
eslint.config.*
commitlint.config.*
.husky/

# OS files
.DS_Store
Thumbs.db

# Dependencies (rebuilt in Docker)
node_modules/

Add npm scripts:

{
  "docker:build": "docker build -t <project-name> .",
  "docker:run": "docker run -p 3000:3000 <project-name>"
}

Suggest Trivy scanning:

# Scan the built image for vulnerabilities
trivy image --severity HIGH,CRITICAL --exit-code 1 <project-name>

Component: health-checks — Health Check Module

Generate a health check module under src/health/ following the project's coding standard (Result pattern, Zod schemas, branded types, readonly, TSDoc).

src/health/schema.ts:

• Zod schema for health check responses
• Status type using as const object (not enum — Rule 3.5):

  const HealthStatus = {
    HEALTHY: 'HEALTHY',
    DEGRADED: 'DEGRADED',
    UNHEALTHY: 'UNHEALTHY',
  } as const;
  type HealthStatusValue = typeof HealthStatus[keyof typeof HealthStatus];
  

• Response types for liveness, readiness (with per-component status), and startup
• All properties readonly (Rule 7.1)
• Exhaustive switch with assertUnreachable (Rule 8.3)

src/health/health.ts:

• checkLiveness(): Returns Result<LivenessResponse>. Checks:

  • Process is running
  • Event loop is responsive (measure event loop lag)
  • Does NOT check downstream dependencies (liveness should not trigger restarts due to external failures)

• checkReadiness(options): Returns Promise<Result<ReadinessResponse>>. Checks each dependency:

  • Accept a ReadonlyArray<HealthDependency> (dependency name + check function)
  • Each check has an individual timeout using AbortController + Promise.race (Rule 4.2)
  • Default timeout: 5 seconds per dependency
  • Returns per-component status with overall status
  • Bounded parallelism: check dependencies concurrently but with a limit (Rule 4.3)

• checkStartup(): Returns Result<StartupResponse>. Validates initialization prerequisites.

• All functions <= 40 lines (Rule 8.4), use Result pattern (Rule 6.2), have TSDoc (Rule 10.1)

src/health/health.test.ts:

• Unit tests for all three health check functions
• Tests for: all healthy, one degraded, one unhealthy, timeout scenario, all down
• Property-based test with fast-check for dependency check arrays
• No any in tests

src/health/index.ts:

• Barrel export of public types and functions

Kubernetes probe configuration:

Generate a docs/kubernetes-probes.yaml reference:

livenessProbe:
  httpGet:
    path: /livez
    port: 3000
  initialDelaySeconds: 30
  periodSeconds: 15
  failureThreshold: 3
  timeoutSeconds: 5
readinessProbe:
  httpGet:
    path: /readyz
    port: 3000
  initialDelaySeconds: 10
  periodSeconds: 10
  failureThreshold: 3
  timeoutSeconds: 5
startupProbe:
  httpGet:
    path: /healthz
    port: 3000
  initialDelaySeconds: 5
  periodSeconds: 5
  failureThreshold: 30
  timeoutSeconds: 5

Integration guidance:

Show how to wire the health module into common frameworks:

• Plain http.createServer: route handler example
• Express: app.get('/healthz', ...) example
• Fastify: route example

Component: compose — Docker Compose

Generate docker-compose.yml for development:

services:
  app:
    build:
      context: .
      target: builder  # Use build stage for development
    ports:
      - "3000:3000"
      - "9229:9229"  # Node.js debugger
    volumes:
      - ./src:/app/src
      - ./dist:/app/dist
    environment:
      - NODE_ENV=development
      - LOG_LEVEL=debug
    command: npm run dev
    # Add dependencies as needed:
    # depends_on:
    #   - postgres
    #   - redis

If the user has databases or caches:

Ask which services to include and add them:

• PostgreSQL: with health check, volume for persistence
• Redis: with health check, memory limits
• MongoDB: with health check, volume

Generate docker-compose.prod.yml (production overrides):

services:
  app:
    build:
      context: .
      target: production
    restart: unless-stopped
    environment:
      - NODE_ENV=production
    deploy:
      resources:
        limits:
          memory: 512M

Summary

After generating all selected components:

1. List all created files
2. Note the Kubernetes probe configuration if health checks were generated
3. Remind the user to:
   • Run docker build to verify the Dockerfile works
   • Run Trivy scan on the built image
   • Wire health check endpoints into their HTTP server
   • Add src/health/ to the parent barrel export if applicable