Power up your AI coding assistants with your own custom knowledge base and task management as an MCP server

Quick Start • Upgrading • What's Included • Architecture • Troubleshooting

Archon is currently in beta! Expect things to not work 100%, and please feel free to share any feedback and contribute with fixes/new features! Thank you to everyone for all the excitement we have for Archon already, as well as the bug reports, PRs, and discussions. It's a lot for our small team to get through but we're committed to addressing everything and making Archon into the best tool it possibly can be!

Archon is the command center for AI coding assistants. For you, it's a sleek interface to manage knowledge, context, and tasks for your projects. For the AI coding assistant(s), it's a Model Context Protocol (MCP) server to collaborate on and leverage the same knowledge, context, and tasks. Connect Claude Code, Kiro, Cursor, Windsurf, etc. to give your AI agents access to:

• Your documentation (crawled websites, uploaded PDFs/docs)
• Smart search capabilities with advanced RAG strategies
• Task management integrated with your knowledge base
• Real-time updates as you add new content and collaborate with your coding assistant on tasks
• Much more coming soon to build Archon into an integrated environment for all context engineering

This new vision for Archon replaces the old one (the agenteer). Archon used to be the AI agent that builds other agents, and now you can use Archon to do that and more.

It doesn't matter what you're building or if it's a new/existing codebase - Archon's knowledge and task management capabilities will improve the output of any AI driven coding.

• GitHub Discussions - Join the conversation and share ideas about Archon
• Contributing Guide - How to get involved and contribute to Archon
• Introduction Video - Getting started guide and vision for Archon
• Archon Kanban Board - Where maintainers are managing issues/features
• Dynamous AI Mastery - The birthplace of Archon - come join a vibrant community of other early AI adopters all helping each other transform their careers and businesses!

• Docker Desktop
• Node.js 18+ (for hybrid development mode)
• Supabase account (free tier or local Supabase both work)
• OpenAI API key (Gemini and Ollama are supported too!)
• (OPTIONAL) Make (see Installing Make below)

1. Clone Repository:

   git clone https://github.com/coleam00/archon.git

   cd archon

2. Environment Configuration:

   cp .env.example .env
   # Edit .env and add your Supabase credentials:
   # SUPABASE_URL=https://your-project.supabase.co
   # SUPABASE_SERVICE_KEY=your-service-key-here

   IMPORTANT NOTES:

   • For cloud Supabase: they recently introduced a new type of service role key but use the legacy one (the longer one).
   • For local Supabase: set SUPABASE_URL to http://host.docker.internal:8000 (unless you have an IP address set up).

3. Database Setup: In your Supabase project SQL Editor, copy, paste, and execute the contents of migration/complete_setup.sql

4. Start Services (choose one):

   Full Docker Mode (Recommended for Normal Archon Usage)

   docker compose up --build -d

   This starts all core microservices in Docker:

   • Server: Core API and business logic (Port: 8181)
   • MCP Server: Protocol interface for AI clients (Port: 8051)
   • UI: Web interface (Port: 3737)

   Ports are configurable in your .env as well!

5. Configure API Keys:

   • Open http://localhost:3737
   • You'll automatically be brought through an onboarding flow to set your API key (OpenAI is default)

Once everything is running:

1. Test Web Crawling: Go to http://localhost:3737 → Knowledge Base → "Crawl Website" → Enter a doc URL (such as https://ai.pydantic.dev/llms-full.txt)
2. Test Document Upload: Knowledge Base → Upload a PDF
3. Test Projects: Projects → Create a new project and add tasks
4. Integrate with your AI coding assistant: MCP Dashboard → Copy connection config for your AI coding assistant

# Option 1: Using Chocolatey
choco install make

# Option 2: Using Scoop
scoop install make

# Option 3: Using WSL2
wsl --install
# Then in WSL: sudo apt-get install make

# Make comes pre-installed on macOS
# If needed: brew install make

# Debian/Ubuntu
sudo apt-get install make

# RHEL/CentOS/Fedora
sudo yum install make

If you need to completely reset your database and start fresh:

To upgrade Archon to the latest version:

1. Pull latest changes:

   git pull

2. Check for migrations: Look in the migration/ folder for any SQL files newer than your last update. Check the file created dates to determine if you need to run them. You can run these in the SQL editor just like you did when you first set up Archon. We are also working on a way to make handling these migrations automatic!

3. Rebuild and restart:

   docker compose up -d --build

This is the same command used for initial setup - it rebuilds containers with the latest code and restarts services.

• Smart Web Crawling: Automatically detects and crawls entire documentation sites, sitemaps, and individual pages
• Document Processing: Upload and process PDFs, Word docs, markdown files, and text documents with intelligent chunking
• Code Example Extraction: Automatically identifies and indexes code examples from documentation for enhanced search
• Vector Search: Advanced semantic search with contextual embeddings for precise knowledge retrieval
• Source Management: Organize knowledge by source, type, and tags for easy filtering

• Model Context Protocol (MCP): Connect any MCP-compatible client (Claude Code, Cursor, even non-AI coding assistants like Claude Desktop)
• MCP Tools: Comprehensive yet simple set of tools for RAG queries, task management, and project operations
• Multi-LLM Support: Works with OpenAI, Ollama, and Google Gemini models
• RAG Strategies: Hybrid search, contextual embeddings, and result reranking for optimal AI responses
• Real-time Streaming: Live responses from AI agents with progress tracking

• Hierarchical Projects: Organize work with projects, features, and tasks in a structured workflow
• AI-Assisted Creation: Generate project requirements and tasks using integrated AI agents
• Document Management: Version-controlled documents with collaborative editing capabilities
• Progress Tracking: Real-time updates and status management across all project activities

• WebSocket Updates: Live progress tracking for crawling, processing, and AI operations
• Multi-user Support: Collaborative knowledge building and project management
• Background Processing: Asynchronous operations that don't block the user interface
• Health Monitoring: Built-in service health checks and automatic reconnection

Archon uses true microservices architecture with clear separation of concerns:

┌─────────────────┐    ┌─────────────────┐    ┌─────────────────┐    ┌─────────────────┐
│   Frontend UI   │    │  Server (API)   │    │   MCP Server    │    │ Agents Service  │
│                 │    │                 │    │                 │    │                 │
│  React + Vite   │◄──►│    FastAPI +    │◄──►│    Lightweight  │◄──►│   PydanticAI    │
│  Port 3737      │    │    SocketIO     │    │    HTTP Wrapper │    │   Port 8052     │
│                 │    │    Port 8181    │    │    Port 8051    │    │                 │
└─────────────────┘    └─────────────────┘    └─────────────────┘    └─────────────────┘
         │                        │                        │                        │
         └────────────────────────┼────────────────────────┼────────────────────────┘
                                  │                        │
                         ┌─────────────────┐               │
                         │    Database     │               │
                         │                 │               │
                         │    Supabase     │◄──────────────┘
                         │    PostgreSQL   │
                         │    PGVector     │
                         └─────────────────┘

• HTTP-based: All inter-service communication uses HTTP APIs
• Socket.IO: Real-time updates from Server to Frontend
• MCP Protocol: AI clients connect to MCP Server via SSE or stdio
• No Direct Imports: Services are truly independent with no shared code dependencies

• Lightweight Containers: Each service contains only required dependencies
• Independent Scaling: Services can be scaled independently based on load
• Development Flexibility: Teams can work on different services without conflicts
• Technology Diversity: Each service uses the best tools for its specific purpose

By default, Archon services run on the following ports:

• archon-ui: 3737
• archon-server: 8181
• archon-mcp: 8051
• archon-agents: 8052
• archon-docs: 3838 (optional)

To use custom ports, add these variables to your .env file:

# Service Ports Configuration
ARCHON_UI_PORT=3737
ARCHON_SERVER_PORT=8181
ARCHON_MCP_PORT=8051
ARCHON_AGENTS_PORT=8052
ARCHON_DOCS_PORT=3838

Example: Running on different ports:

ARCHON_SERVER_PORT=8282
ARCHON_MCP_PORT=8151

By default, Archon uses localhost as the hostname. You can configure a custom hostname or IP address by setting the HOST variable in your .env file:

# Hostname Configuration
HOST=localhost  # Default

# Examples of custom hostnames:
HOST=192.168.1.100     # Use specific IP address
HOST=archon.local      # Use custom domain
HOST=myserver.com      # Use public domain

This is useful when:

• Running Archon on a different machine and accessing it remotely
• Using a custom domain name for your installation
• Deploying in a network environment where localhost isn't accessible

After changing hostname or ports:

1. Restart Docker containers: docker compose down && docker compose --profile full up -d
2. Access the UI at: http://${HOST}:${ARCHON_UI_PORT}
3. Update your AI client configuration with the new hostname and MCP port

# Install dependencies
make install

# Start development (recommended)
make dev        # Backend in Docker, frontend local with hot reload

# Alternative: Everything in Docker
make dev-docker # All services in Docker

# Stop everything (local FE needs to be stopped manually)
make stop

Best for active development with instant frontend updates:

• Backend services run in Docker (isolated, consistent)
• Frontend runs locally with hot module replacement
• Instant UI updates without Docker rebuilds

For all services in Docker environment:

• All services run in Docker containers
• Better for integration testing
• Slower frontend updates

# Run tests
make test       # Run all tests
make test-fe    # Run frontend tests
make test-be    # Run backend tests

# Run linters
make lint       # Lint all code
make lint-fe    # Lint frontend code
make lint-be    # Lint backend code

# Check environment
make check      # Verify environment setup

# Clean up
make clean      # Remove containers and volumes (asks for confirmation)

# View logs using Docker Compose directly
docker compose logs -f              # All services
docker compose logs -f archon-server # API server
docker compose logs -f archon-mcp    # MCP server
docker compose logs -f archon-ui     # Frontend

Note: The backend services are configured with --reload flag in their uvicorn commands and have source code mounted as volumes for automatic hot reloading when you make changes.

If you see "Port already in use" errors:

# Check what's using a port (e.g., 3737)
lsof -i :3737

# Stop all containers and local services
make stop

# Change the port in .env

If you encounter permission errors with Docker:

# Add your user to the docker group
sudo usermod -aG docker $USER

# Log out and back in, or run
newgrp docker

• Make not found: Install Make via Chocolatey, Scoop, or WSL2 (see Installing Make)
• Line ending issues: Configure Git to use LF endings:

  git config --global core.autocrlf false

• Check backend is running: curl http://localhost:8181/health
• Verify port configuration in .env
• For custom ports, ensure both ARCHON_SERVER_PORT and VITE_ARCHON_SERVER_PORT are set

If docker compose commands hang:

# Reset Docker Compose
docker compose down --remove-orphans
docker system prune -f

# Restart Docker Desktop (if applicable)

• Frontend: Ensure you're running in hybrid mode (make dev) for best HMR experience
• Backend: Check that volumes are mounted correctly in docker-compose.yml
• File permissions: On some systems, mounted volumes may have permission issues

Archon Community License (ACL) v1.2 - see LICENSE file for details.

TL;DR: Archon is free, open, and hackable. Run it, fork it, share it - just don't sell it as-a-service without permission.