Aiops Gordon

Docker AI (Gordon) Setup Skill

Quick Start

1. Verify Docker Desktop - Version 4.53+ required
2. Enable Gordon - Settings → Beta features → Toggle "Docker AI"
3. Test Gordon - Run docker ai "What can you do?"
4. Use Gordon - For building, debugging, and optimizing containers

What is Docker AI (Gordon)?

Docker AI (codenamed Gordon) is an AI-powered assistant built into Docker Desktop that helps with:

• Building containers - Generate Dockerfiles from requirements
• Debugging issues - Analyze logs and errors
• Optimizing images - Reduce size and improve performance
• Scanning vulnerabilities - Security analysis
• Explaining commands - Learn Docker and Kubernetes

Enabling Docker AI

Docker Desktop 4.53+

1. Open Docker Desktop
2. Go to Settings → Beta features
3. Toggle on "Docker AI"
4. Click "Get started" or "Enable"
5. Accept terms and conditions

Alternative: Docker CLI Extension

# If not using Docker Desktop, try CLI extension
# (Note: Gordon is primarily in Docker Desktop)
docker context list
# See if AI context is available

Gordon Commands Reference

Building Containers

# Generate Dockerfile from requirements
docker ai "Create a Dockerfile for FastAPI application with Python 3.13"

# Optimize existing Dockerfile
docker ai "Optimize this Dockerfile for smaller image size" < Dockerfile

# Create multi-stage Dockerfile
docker ai "Convert this to a multi-stage Dockerfile" < Dockerfile

# Generate docker-compose.yml
docker ai "Create a docker-compose file with frontend, backend, and mcp-server services"

Debugging Issues

# Analyze startup failures
docker ai "Why is my container exiting immediately?"

# Debug build errors
docker ai "Fix this Docker build error" < error.log

# Troubleshoot networking
docker ai "Why can't container1 connect to container2?"

# Analyze logs
docker ai "Analyze these container logs for issues" < logs.txt

Image Optimization

# Reduce image size
docker ai "Make this Dockerfile produce a smaller image"

# Optimize layer caching
docker ai "Optimize layer caching in this Dockerfile"

# Security scan
docker ai "Scan this image for security vulnerabilities" todo-backend:latest

# Best practices review
docker ai "Review this Dockerfile for security best practices"

Multi-Service Orchestration

# Generate compose file
docker ai "Create docker-compose for Next.js frontend, FastAPI backend, PostgreSQL"

# Add health checks
docker ai "Add health checks to this docker-compose file"

# Configure networking
docker ai "Set up bridge networking for these 3 containers"

Gordon Examples for Todo Project

Frontend Dockerfile Generation

# Ask Gordon to create Next.js Dockerfile
docker ai "Create a production-ready Dockerfile for Next.js 16 with TypeScript, using multi-stage build, with nginx or standalone server"

Expected output from Gordon:

# Gordon will generate optimized Dockerfile
# Review and save to frontend/Dockerfile

Backend Dockerfile Generation

docker ai "Create a multi-stage Dockerfile for FastAPI with Python 3.13, SQLModel, and OpenAI dependencies"

MCP Server Dockerfile

docker ai "Create a minimal Dockerfile for Python FastMCP server with minimal dependencies"

Docker Compose Generation

docker ai "Create a docker-compose.yml file for:
- Frontend: Next.js on port 3000
- Backend: FastAPI on port 8000
- MCP Server: Python on port 8001
- Network: todo-network
- Health checks for all services
- Environment variables from .env file"

Debug Build Issues

# If build fails, paste the error
docker ai "I'm getting this error when building:
Error: ModuleNotFoundError: No module named 'fastapi'
How do I fix this?"

# Or redirect error file
docker ai "Fix the errors in this build log" < build-error.log

Gordon Limitations

Gordon vs Manual Docker

Best Practices with Gordon

1. Always review output - Gordon generates good code, but review before saving
2. Iterative refinement - Use follow-up prompts to refine Gordon's output
3. Test locally - Always test Dockerfiles generated by Gordon
4. Version control - Save Gordon's suggestions and iterate
5. Learn from Gordon - Understand why Gordon suggests certain patterns

Integration with Other Skills

Combine Gordon with:

• @docker-containerization-builder - Use Gordon for Dockerfile, agent for full setup
• @devops-kubernetes-builder - Gordon helps debug K8s container issues
• @aiops-helm-builder - Use Gordon for chart image optimization

Gordon Troubleshooting

Example Gordon Workflows

Workflow 1: Generate Dockerfile

# Step 1: Ask Gordon
docker ai "Create Dockerfile for Next.js with TypeScript and Tailwind CSS"

# Step 2: Save output to file
# Copy Gordon's response to Dockerfile

# Step 3: Build and test
docker build -t test-image .
docker run --rm test-image

Workflow 2: Debug Container Issue

# Step 1: Get container logs
docker logs <container-name> > logs.txt

# Step 2: Ask Gordon
docker ai "Analyze this log file and tell me what's wrong" < logs.txt

# Step 3: Apply fix
# Update Dockerfile or code based on Gordon's suggestion

# Step 4: Rebuild and test
docker build -t fixed-image .
docker run --rm fixed-image

Workflow 3: Optimize Production Image

# Step 1: Build initial image
docker build -t todo-backend:latest ./backend

# Step 2: Check size
docker images todo-backend

# Step 3: Ask Gordon to optimize
docker ai "Reduce the size of this Dockerfile while maintaining all functionality" < backend/Dockerfile

# Step 4: Compare
docker build -t todo-backend:optimized ./backend
docker images | grep todo-backend

Verification Checklist

After using Gordon:

• Gordon output reviewed and understood
• Generated files are saved with correct names
• Dockerfiles build without errors
• Containers start and run correctly
• Health checks pass
• Services can communicate
• Image size is reasonable
• Security scan shows no critical vulnerabilities

Next Steps

After successful Docker setup with Gordon:

1. Test with docker-compose up -d
2. Load images to Minikube
3. Deploy with Kubernetes manifests
4. Convert to Helm charts

References

• Docker AI Documentation
• Docker Desktop Release Notes
• Docker Best Practices
• Phase 4 Constitution
• Phase 4 Plan