Docker Helper

Simplify Docker container management, compose setup, and debugging through natural language commands.

When to Use

✅ USE this skill when:

• "Show me all running containers"
• "Start/stop/restart container [name]"
• "Show logs for [container]"
• "Check disk usage of Docker"
• "Prune unused containers/images"
• "Create a docker-compose.yml for [service]"
• "What's running on port 8080?"
• "Container [name] won't start, debug it"

❌ DON'T use this skill when:

• Need Kubernetes management → use a K8s skill
• Building complex multi-service orchestrations → consider Docker Swarm/K8s skills
• Production deployment with zero-downtime → use deployment-specific skills

Workflow

1. Container Management

When the user asks about containers, the agent runs:

# List all
docker ps -a
docker ps --format "table {{.Names}}\t{{.Status}}\t{{.Ports}}"

# Inspect specific
docker inspect [container] | jq '.[0].NetworkSettings.Ports'
docker stats [container] --no-stream

# Control
docker start/stop/restart [container]
docker rm -f [container]    (only with user confirmation)

2. Logs & Debugging

For debugging container issues:

# Standard logs
docker logs [container] --tail 50
docker logs [container] --tail 200 | grep -i error

# Resource issues
docker stats --no-stream

# Check why container exited
docker inspect [container] | jq '.[0].State'
docker logs [container] --tail 20 --timestamps

3. Image Management

# List images
docker images

# Build
docker build -t [name]:[tag] .

# Cleanup (with confirmation)
docker image prune -a -f

4. Docker Compose

When the user needs a compose file:

1. Ask what services they need (e.g., "nginx + postgres + redis")
2. Generate a docker-compose.yml with:
• Appropriate image tags (avoid latest in production)
• Volume mounts for persistence
• Network configuration
• Health checks where applicable
• Environment variable stubs (.env file)
3. Show the user the generated file, offer to adjust

5. Resource Cleanup (WITH CONFIRMATION)

Always ask for confirmation before:

• docker system prune -a -f (removes ALL unused resources)
• docker volume prune -f (removes orphaned volumes with data)
• docker image rm without the -f flag first

Common Templates

Web App (Node/Python + Postgres):

services:
  app:
    build: .
    ports: ["${PORT}:${PORT}"]
    env_file: .env
    depends_on: [db]
  db:
    image: postgres:16-alpine
    volumes: ["pgdata:/var/lib/postgresql/data"]
    environment:
      POSTGRES_DB: ${DB_NAME}
      POSTGRES_PASSWORD: ${DB_PASS}
volumes: {pgdata:}

Nginx Reverse Proxy:

services:
  nginx:
    image: nginx:alpine
    ports: ["80:80", "443:443"]
    volumes: ["./nginx.conf:/etc/nginx/nginx.conf:ro", "./html:/usr/share/nginx/html:ro"]

Examples

> User: "What containers are running?"

> Agent: Runs docker ps --format "table..." and shows a clean table

>

> User: "My nginx container won't start"

> Agent: Checks logs + inspect, identifies the port conflict or config error

>

> User: "Clean up Docker, it's using too much space"

> Agent: Runs docker system df, shows usage, asks what to prune

>

> User: "Create a compose file for a Python app with Redis"

> Agent: Generates docker-compose.yml + .env.example, explains each part

Notes

• Always prefix destructive commands with confirmation
• Use docker compose (v2) over docker-compose (v1) when available
• Prefer alpine-based images for smaller footprint
• For security: never expose Docker socket without TLS