DockAI 🐳🤖

The Customizable AI Dockerfile Generation Framework

DockAI is an intelligent, adaptive framework that generates production-ready Dockerfiles for any project using Large Language Models (LLMs). It goes beyond simple template generation by understanding your codebase through RAG (Retrieval-Augmented Generation), analyzing your project architecture, and iteratively improving Dockerfiles until they pass all security and validation checks.

🌟 Key Features

🧠 Intelligent Context Understanding

• RAG-Powered Analysis: Uses semantic embeddings (sentence-transformers) to understand your entire codebase
• AST Code Intelligence: Extracts entry points, ports, environment variables, and framework dependencies automatically
• Multi-Language Support: Works with JavaScript/TypeScript, Python, Go, Java, Ruby, PHP, .NET, and more

🏗️ Multi-Agent Architecture

DockAI v4.0 features a sophisticated multi-agent system orchestrated by LangGraph:

• Analyzer Agent: Project discovery and technology stack detection
• Blueprint Agent: Architectural planning and runtime configuration
• Generator Agent: Dockerfile creation with best practices
• Iterative Generator Agent: Refining existing Dockerfiles based on feedback
• Reviewer Agent: Security auditing and vulnerability detection
• Reflector Agent: Failure analysis and adaptive learning
• Error Analyzer Agent: Classification of build/runtime errors for better recovery
• Iterative Improver Agent: Surgical fixes based on validation feedback

🔄 Adaptive & Self-Improving

• Automatic Validation: Builds and tests the Docker image locally
• Iterative Refinement: Learns from failures and auto-fixes issues (up to configurable retries)
• Smart Fallback: Reverts to the last working Dockerfile if fixes fail (ignoring non-critical warnings)
• Smart Reflection: AI analyzes build/runtime errors and adjusts strategy
• Reanalysis: Detects when fundamental assumptions are wrong and pivots

🔒 Security & Best Practices

• Hadolint Integration: Strict Dockerfile linting (warnings are treated as errors and auto-fixed)
• Trivy Security Scanning: Container vulnerability detection
• AI Security Review: Identifies security anti-patterns (root users, exposed secrets, etc.)
• Multi-Stage Builds: Optimizes for smaller, more secure images

🎯 Production-Ready Features

• Health Check Detection: Auto-discovers and configures health endpoints
• Resource Optimization: Configurable memory, CPU, and process limits
• Multi-Platform Support: Works with Docker, Podman, and GitHub Actions
• Observability: OpenTelemetry and LangSmith tracing support

🛠️ Highly Customizable

• Multi-LLM Support: OpenAI, Google Gemini, Anthropic Claude, Azure OpenAI, Ollama
• Per-Agent Model Selection: Choose different models for different tasks (cost vs. quality)
• Custom Instructions: Override default agent behavior
• Custom Prompts: Complete control over AI reasoning
• Environment-Based Configuration: 100+ configuration options via environment variables

🏛️ Architecture

DockAI v4.0 is built on a modern, agent-based architecture using LangGraph for workflow orchestration:

graph TD
    Start([Start]) --> Scan[Scan Repo]
    Scan --> Analyze[Agent : Analyzer]
    Analyze --> ReadFiles[Read Context]
    ReadFiles --> Blueprint[Agent : Blueprint]
    Blueprint --> Generate[Agent : Generator]
    Generate --> Review[Agent : Reviewer]
    
    Review -->|Secure| Validate[Agent : Validator]
    Review -->|Insecure| CheckRetry{Retry < Max?}
    
    Validate -->|Success| End([Success])
    Validate -->|Fail| CheckRetry
    
    CheckRetry -->|Yes| Reflect[Agent : Reflector]
    CheckRetry -->|No| EndFail([Fail])
    
    Reflect --> IncRetry[Increment Retry]
    IncRetry --> Route{Route Fix}
    
    Route -->|Re-Analyze| Analyze
    Route -->|Re-Plan| Blueprint
    Route -->|Fix Code| Improver[Agent : Iterative Improver]
    Improver --> Review

Loading

Core Components

• LangGraph Workflow Engine: Orchestrates the agent flow with conditional routing
• RAG Context Engine: In-memory vector store for semantic code search
• Multi-Agent System: 8 specialized AI agents for different tasks
• Validation Pipeline: Docker build, Hadolint, Trivy, and health checks
• State Management: Centralized state for workflow coordination

For detailed architecture documentation, see docs/architecture.md.

🎯 Three Ways to Use DockAI

DockAI can be integrated into your workflow in multiple ways, depending on your needs:

1️⃣ CLI Tool (Local Development)

Install via pip or uv and use directly from the command line:

Using pip:

pip install dockai-cli
export OPENAI_API_KEY="your-key"
dockai build .

Using uv (faster):

# Install uv
curl -LsSf https://astral.sh/uv/install.sh | sh

# Install DockAI
uv pip install dockai-cli

# Use it
dockai build .

Perfect for:

• Local development and testing
• Quick Dockerfile generation
• Iterating on containerization

2️⃣ GitHub Actions (CI/CD Automation)

Integrate DockAI into your CI/CD pipeline with the GitHub Action:

name: Generate Dockerfile
on: [push]

jobs:
  dockerize:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      
      - name: Generate Dockerfile with DockAI
        uses: itzzjb/dockai@v4
        with:
          openai_api_key: ${{ secrets.OPENAI_API_KEY }}
          project_path: '.'
      
      - name: Commit Dockerfile
        run: |
          git config user.name "DockAI Bot"
          git add Dockerfile
          git commit -m "chore: update Dockerfile"
          git push

Perfect for:

• Automated Dockerfile updates
• Multi-service monorepos
• Continuous integration workflows
• Team collaboration

Learn more: GitHub Actions Guide

3️⃣ MCP Integration (Model Context Protocol)

Use DockAI as an MCP server with AI assistants like Claude Desktop:

Setup MCP:

{
  "mcpServers": {
    "dockai": {
      "command": "uvx",
      "args": ["dockai-cli"]
    }
  }
}

Usage with Claude Desktop:

You: Can you dockerize this Node.js project?
Claude: [Uses DockAI MCP] I'll generate a Dockerfile for your project...

Perfect for:

• Interactive AI-assisted development
• Natural language Dockerfile generation
• Integration with Claude Desktop, VSCode, and other MCP clients
• Conversational containerization workflow

Learn more: MCP Integration Guide

---

🚀 Quick Start

Prerequisites

• Python 3.10 or higher
• Docker installed and running
• An API key from at least one LLM provider (OpenAI, Google, Anthropic, etc.)

Installation

Option 1: Install from PyPI (Recommended)

pip install dockai-cli

Option 2: Install with uv (Faster)

uv is an extremely fast Python package installer and resolver:

# Install uv (if not already installed)
curl -LsSf https://astral.sh/uv/install.sh | sh

# Install DockAI
uv pip install dockai-cli

Why use uv?

• 🚀 10-100x faster than pip
• 📦 Drop-in replacement for pip
• 🔒 Deterministic dependency resolution

Option 3: Install from Source

git clone https://github.com/itzzjb/dockai.git
cd dockai
pip install -e .

Basic Usage

1. Set up your API key (choose one provider):

# OpenAI (Default)
export OPENAI_API_KEY="your-api-key-here"

# Or Google Gemini
export GOOGLE_API_KEY="your-api-key-here"
export DOCKAI_LLM_PROVIDER="gemini"

# Or Anthropic Claude
export ANTHROPIC_API_KEY="your-api-key-here"
export DOCKAI_LLM_PROVIDER="anthropic"

2. Navigate to your project and run DockAI:

cd /path/to/your/project
dockai build .

3. Done! Your production-ready Dockerfile will be created and validated.

Example Output

🔍 Scanning project...
✓ Found 42 files

🧠 Analyzing project with AI...
✓ Detected: Node.js Express application
✓ Entry point: src/server.js
✓ Dependencies: package.json

📖 Reading files with RAG (10 relevant chunks)...
✓ Context retrieved

🏗️ Creating architectural blueprint...
✓ Multi-stage build planned
✓ Health endpoint: /health

🔨 Generating Dockerfile...
✓ Dockerfile created

🔍 Reviewing security...
✓ No critical issues found

🧪 Validating with Docker...
✓ Image built successfully (142 MB)
✓ Container started
✓ Health check passed

✅ Dockerfile generated successfully!

📚 Documentation

Comprehensive documentation is available in the docs/ directory:

• Getting Started - Installation, setup, and first steps
• Architecture - Deep dive into v4.0 architecture and RAG system
• Configuration - All environment variables and customization options
• LLM Providers - Setting up different LLM providers
• GitHub Actions - Using DockAI in CI/CD pipelines
• MCP Integration - Using DockAI with Model Context Protocol
• API Reference - Code-level documentation
• FAQ - Common questions and troubleshooting

🎯 Use Cases

Single Project Dockerization

# Automatically detects and handles any project type
cd /path/to/your/project
dockai build .

Polyglot Projects

# Works with multi-language projects
dockai build ./my-fullstack-app

Microservices Architecture

# Generate optimized Dockerfiles for each service
for service in api frontend worker; do
  dockai build ./services/$service
done

Custom Requirements

# Add specific requirements for your organization
export DOCKAI_GENERATOR_INSTRUCTIONS="Always use Alpine Linux and pin all versions. Include MAINTAINER label."
dockai build .

Cost-Optimized Generation

# Use cheaper models for analysis, powerful models for generation
export DOCKAI_MODEL_ANALYZER="gpt-4o-mini"
export DOCKAI_MODEL_GENERATOR="gpt-4o"
dockai build .

⚙️ Configuration

DockAI offers extensive configuration through environment variables:

Model Selection

# Use different models for different agents (cost optimization)
export DOCKAI_MODEL_ANALYZER="gpt-4o-mini"      # Fast, cheap model for analysis
export DOCKAI_MODEL_GENERATOR="gpt-4o"          # Powerful model for generation
export DOCKAI_MODEL_REFLECTOR="gemini-1.5-pro"  # Strong reasoning for failure analysis

Security & Validation

export DOCKAI_SKIP_HADOLINT="false"            # Enable Dockerfile linting
export DOCKAI_SKIP_SECURITY_SCAN="false"       # Enable Trivy scanning
export DOCKAI_STRICT_SECURITY="true"           # Fail on any vulnerability
export DOCKAI_MAX_IMAGE_SIZE_MB="500"          # Max acceptable image size

RAG Configuration

export DOCKAI_USE_RAG="true"                   # Enable RAG (default in v4.0)
export DOCKAI_EMBEDDING_MODEL="all-MiniLM-L6-v2"  # Embedding model
export DOCKAI_READ_ALL_FILES="true"            # Read all source files

Retry & Adaptation

export MAX_RETRIES="3"                         # Max attempts to fix failures

For complete configuration options, see docs/configuration.md.

🧪 Testing

Run the test suite:

# Install test dependencies
pip install -e ".[test]"

# Run all tests
pytest

# Run with coverage
pytest --cov=src/dockai --cov-report=html

🤝 Contributing

Contributions are welcome! Please see our Contributing Guide for details.

🗺️ Roadmap

• Support for Docker Compose generation
• .dockerignore file generation
• Multi-stage build optimization advisor
• Integration with container registries
• Web UI for interactive generation
• Plugin system for custom validators

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• LangGraph for the agent orchestration framework
• LangChain for LLM abstractions and tools
• Sentence Transformers for efficient embeddings
• All the open-source projects that make DockAI possible

📧 Support

• Issues: GitHub Issues
• Discussions: GitHub Discussions
• Email: desilvabethmin@gmail.com

---

Made with ❤️ by Januda Bethmin

⭐ If you find DockAI useful, please give it a star on GitHub!