mcp -python

This MCP server, mcp-server-python, runs via uv to provide language-model-enabled retrieval augmented generation (RAG) capabilities backed by Inkeep content. It authenticates with your Inkeep API key and uses the Inkeep RAG model to fetch product documentation and related content on demand. To operate it, you first ensure uv and Python dependencies are installed, then point uv at the local project path so it can load the inkeep_mcp_server module. The server is configured through environment variables that specify the API base URL, API key, the model to use, and the tool metadata for Claude Desktop or other MCP clients. Once running, clients can query the MCP server to obtain structured product content, summaries, and guidance derived from Inkeep documentation and product pages.

• Prerequisites:

  • Python installed and a supported environment
  • uv (Python project manager) installed
  • Access to an Inkeep account with an API key

• Local setup:

  1. Clone the repository: git clone https://github.com/inkeep/mcp-server-python.git
  2. Navigate into the project: cd mcp-server-python
  3. Set up the Python environment with uv and install dependencies: uv venv uv pip install -r pyproject.toml
  4. Note the absolute path to the project directory. You will use it to configure the MCP client (via the mcp_config snippet shown in this document).
  5. Obtain an Inkeep API key as described in the README and prepare your environment variables as shown in the mcp_config example.

• Running the server:

  • Ensure you are in the project root and have the environment set up.
  • Use the following mcp_config entry as part of your MCP client configuration to launch the server via uv: { "command": "uv", "args": [ "--directory", "<YOUR_INKEEP_MCP_SERVER_ABSOLUTE_PATH>", "run", "-m", "inkeep_mcp_server" ], "env": { "INKEEP_API_BASE_URL": "https://api.inkeep.com/v1", "INKEEP_API_KEY": "<YOUR_INKEEP_API_KEY>", "INKEEP_API_MODEL": "inkeep-rag", "INKEEP_MCP_TOOL_NAME": "search-product-content", "INKEEP_MCP_TOOL_DESCRIPTION": "Retrieves product documentation about Inkeep. The query should be framed as a conversational question about Inkeep." } }
  • Run the MCP client with this configuration and start issuing queries through your preferred interface (e.g., Claude Desktop or another MCP frontend).

• Notes:

  • The exact module name to run may depend on how you package the server; the example uses inkeep_mcp_server as the module.
  • If uv is installed in a virtual environment, ensure the path provided in --directory points to the project root so uv can locate the module.
  • Keep your API key secure and do not commit it to source control.

Tips and common issues:

• Ensure the Inkeep API key has the required permissions for content retrieval and RAG processing.
• If uv cannot be found, verify your PATH or provide the full path to the uv executable in your MCP client configuration.
• When changing environment variables, restart the MCP client to apply updates.
• If you encounter authentication errors, double-check INKEEP_API_KEY and INKEEP_API_BASE_URL values.
• For debugging, you can run the module directly with Python if needed, but uv is the recommended workflow per the README.
• Remember to replace placeholders like <YOUR_INKEEP_MCP_SERVER_ABSOLUTE_PATH> and <YOUR_INKEEP_API_KEY> with real values before starting the server.