NOTE: This project is no longer being maintained, as issues with model memory for smaller models means my preferred method for MCP servers is as individual containerized server providers rather than a monolithic routing provider. Code is being kept up in case others still want to try to go down this route.

- MCP Gateway: A unified endpoint for all tool requests following the MCP standard
- MCP Manifest: An endpoint that describes all available tools and their capabilities
- Direct Tool Access: Each tool can be accessed directly via its own API endpoints
- Modular Design: Easy to add or remove tools as needed

1. GitHub Tool: Interact with GitHub repositories, issues, and search
2. GitLab Tool: Interact with GitLab projects, issues, and pipelines
3. Google Maps Tool: Geocoding, directions, and places search
4. Memory Tool: Store and retrieve data persistently
5. Puppeteer Tool: Take screenshots, generate PDFs, and extract content from websites

   git clone https://github.com/yourusername/mcp-server.git

   cd mcp-server

   

   pip install -r requirements.txt

   

   npm install

   

   SECRET_KEY=your-secret-key

   DEBUG=False

   

   # GitHub configuration

   GITHUB_TOKEN=your-github-token

   

   # GitLab configuration

   GITLAB_TOKEN=your-gitlab-token

   

   # Google Maps configuration

   GMAPS_API_KEY=your-google-maps-api-key

   

   # Memory configuration

   MEMORY_DB_URI=sqlite:///memory.db

   

   # Puppeteer configuration

   PUPPETEER_HEADLESS=true

   CHROME_PATH=/usr/bin/chromium-browser

   

   python app.py

   

   docker build -t mcp-server .

   

   docker run -p 5000:5000 --env-file .env mcp-server

   

   version: '3'

   services:

     mcp-server:

       build: .

       ports:

         - "5000:5000"

       volumes:

         - ./data:/app/data

       env_file:

         - .env

       restart: unless-stopped

   

   docker-compose up -d

   

   podman build -t mcp-server .

   

   podman run -p 5000:5000 --env-file .env mcp-server

   

   mkdir -p ./data

   podman run -p 5000:5000 --env-file .env -v ./data:/app/data:Z mcp-server

   

   # Install podman-compose if needed

   pip install podman-compose

   

   # Use the same docker-compose.yml file as above

   podman-compose up -d

   

Endpoint: POST /mcp/gateway

Request format:

{

  "tool": "github",

  "action": "listRepos",

  "parameters": {

    "username": "octocat"

  }

}

Response format:

{

  "tool": "github",

  "action": "listRepos",

  "status": "success",

  "result": [

    {

      "id": 1296269,

      "name": "Hello-World",

      "full_name": "octocat/Hello-World",

      "owner": {

        "login": "octocat",

        "id": 1

      },

      ...

    }

  ]

}

Endpoint: GET /mcp/manifest

Response format:

{

  "manifestVersion": "1.0",

  "tools": {

    "github": {

      "actions": {

        "listRepos": {

          "description": "List repositories for a user or organization",

          "parameters": {

            "username": {

              "type": "string",

              "description": "GitHub username or organization name"

            }

          },

          "returns": {

            "type": "array",

            "description": "List of repository objects"

          }

        },

        ...

      }

    },

    ...

  }

}

Actions:
- listRepos: List repositories for a user or organization
- getRepo: Get details for a specific repository
- searchRepos: Search for repositories
- getIssues: Get issues for a repository
- createIssue: Create a new issue in a repository

Actions:
- listProjects: List all projects accessible by the authenticated user
- getProject: Get details for a specific project
- searchProjects: Search for projects on GitLab
- getIssues: Get issues for a project
- createIssue: Create a new issue in a project
- getPipelines: Get pipelines for a project

Actions:
- geocode: Convert an address to geographic coordinates
- reverseGeocode: Convert geographic coordinates to an address
- getDirections: Get directions between two locations
- searchPlaces: Search for places using the Google Places API
- getPlaceDetails: Get details for a specific place

Actions:
- get: Get a memory item by key
- set: Create or update a memory item
- delete: Delete a memory item by key
- list: List all memory items, with optional filtering
- search: Search memory items by value

Actions**:
- screenshot: Take a screenshot of a webpage
- pdf: Generate a PDF of a webpage
- extract: Extract content from a webpage

Contributing

Contributions are welcome! Here's how you can extend the MCP server:

Adding a New Tool

1. Create a new file in the tools directory, e.g., tools/newtool_tool.py
2. Implement the tool with actions following the same pattern as existing tools
3. Add the tool to the manifest in app.py
4. Register the tool's blueprint in tools/__init__.py

License

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

Acknowledgements

- Model Context Protocol for the standard specification
- Flask for the web framework
- Puppeteer for web automation