Axon.MCP.Server

Axon.MCP.Server exposes a set of MCP tools that let AI assistants like Cursor IDE, Google AntiGravity, and Claude query and analyze your codebase. The server runs as an MCP endpoint and provides semantic search, call graphs, API endpoint discovery, entity mappings, architecture diagrams, and symbol-level details. You can invoke tools such as search, get_call_graph, get_api_endpoints, get_ef_entities, get_module_summary, explore_service, find_implementations, and get_system_architecture_map to extract meaningful context from your repository. Operators can query the server for code meaning, relationships, and architectural metadata, enabling richer, context-aware interactions with AI copilots and IDE integrations.

Prerequisites:

• Python 3.11+ installed on your system
• Access to a PostgreSQL 15+ database and a Redis instance
• Optional: Docker if you prefer containerized deployment

1. Create a Python virtual environment (recommended)

python -m venv venv
source venv/bin/activate  # macOS/Linux
venv\Scripts\activate     # Windows

2. Install dependencies

# If a requirements file is present
pip install -r requirements.txt
# Or install directly from package if published
pip install axon-mcp-server  # replace with actual package name if different

3. Configure environment variables

export DATABASE_URL=postgresql://USER:PASS@HOST:5432/DB_NAME
export REDIS_URL=redis://:PASS@HOST:6379/0
export JWT_PUBLIC_KEY=$(cat path/to/public.pem)
export JWT_PRIVATE_KEY=$(cat path/to/private.pem)

4. Run the MCP server

python -m axon_mcp_server

Alternative (Docker):

# docker run example (adjust image and env vars as needed)
docker run -it --env DATABASE_URL='postgresql://USER:PASS@HOST:5432/DB_NAME' \
  --env REDIS_URL='redis://:PASS@HOST:6379/0' \
  --env JWT_PUBLIC_KEY='...' --env JWT_PRIVATE_KEY='...' \
  ghcr.io/ali-kamali/axon-mcp-server:latest

5. Verify the server is running by hitting the MCP endpoint (default port 8001) or checking logs for readiness.

Notes & tips:

• Ensure PostgreSQL and Redis are reachable from the MCP server host.
• Set appropriate JWT keys for authentication when enabled by your deployment.
• If you enable RBAC or audit logging, configure roles/users in your environment.
• The MCP endpoint typically exposes a REST interface for tool invocations; consult the docs for exact routes and payload formats.
• For large codebases, adjust worker concurrency and database connection pools according to your hardware.
• If using Docker, consider docker-compose to manage service dependencies (db, redis, mcps).