multi-service

This MCP server acts as a modular gateway that exposes a set of tools through the Model Context Protocol. It aggregates GitHub, GitLab, Google Maps, Memory storage, and Puppeteer capabilities behind a unified MCP gateway and a manifest describing each tool. You can interact with the gateway by sending requests to /mcp/gateway using the MCP format, or directly access individual tools via /tool/{tool-name}/... endpoints. The manifest at /mcp/manifest lists available actions for each tool and their required parameters, enabling you to discover capabilities programmatically. This setup lets you route requests to the specific tooling you need (for example GitHub repository operations or memory storage) without building separate clients, while still allowing direct tool access when required.

Prerequisites:

• Python 3.8 or higher
• Node.js 14 or higher (for any optional frontend tooling or script integrations)
• A Linux, macOS, or compatible environment

Installation steps:

1. Clone the repository: git clone https://github.com/yourusername/mcp-server.git cd mcp-server

2. Create and populate environment configuration (see .env guidance in README): SECRET_KEY=your-secret-key DEBUG=False GITHUB_TOKEN=your-github-token GITLAB_TOKEN=your-gitlab-token GMAPS_API_KEY=your-google-maps-api-key MEMORY_DB_URI=sqlite:///memory.db PUPPETEER_HEADLESS=true CHROME_PATH=/usr/bin/chromium-browser

3. Install Python dependencies: pip install -r requirements.txt

4. Install Node.js dependencies (if using any Node-based tooling in the repo): npm install

5. Run the server locally: python app.py

Containerized deployment (optional):

• Build and run with Docker/Docker Compose as described in the repository’s guidance: docker build -t mcp-server . docker run -p 5000:5000 --env-file .env mcp-server or use docker-compose up -d with a docker-compose.yml that mirrors the environment file and volumes.

Notes:

• The project emphasizes a monolithic MCP router approach; if memory constraints apply to smaller models, consider running individual containerized tool servers instead of a single gateway.

Tips and caveats:

• This project is no longer actively maintained; memory management and scaling may require adapting the architecture to individual tool containers.
• Ensure environment variables are securely managed, especially tokens and secrets.
• If you encounter issues with Puppeteer, verify that the Chrome/Chromium path and headless settings are correct for your environment.
• For SELinux-enabled systems, persistent storage using Podman may require volume flags like :Z.
• The MCP manifest and gateway endpoints are the primary surface for discovery and routing; use the gateway to explore capabilities before using direct tool endpoints.