MCP Integration

CiteKit implements the Model Context Protocol (MCP), a standard open protocol that enables AI assistants to connect to data sources.

Because CiteKit is MCP-compliant, it works with any MCP client, including:

• Claude Desktop
• Cline (VS Code Extension)
• Coming soon: JetBrains AI, Zed, and more.

Installation & Setup

You do not need to install a separate "MCP Extension". The citekit package is the server.

1. Install CiteKit

Ensure you have CiteKit installed on your system.

# Python (Recommended)
pip install citekit

# Node.js
npm install citekit

2. Configure Your Client

Claude Desktop

Edit your configuration file:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

Add the following:

Python

{
  "mcpServers": {
    "citekit": {
      "command": "citekit",
      "args": ["serve"],
      "env": {
        "GEMINI_API_KEY": "AIzaSy..."
      }
    }
  }
}

Node.js

{
  "mcpServers": {
    "citekit": {
      "command": "npx",
      "args": ["-y", "citekit", "serve"],
      "env": {
        "GEMINI_API_KEY": "AIzaSy..."
      }
    }
  }
}

Note: If citekit is not in your PATH, provide the full path to your python executable or use uv run citekit serve.

Cline (VS Code)

1. Open Cline Settings.
2. Go to MCP Servers.
3. Add a new server with:
   • Name: citekit
   • Command: citekit (or full path)
   • Args: serve
   • Environment Variables: {"GEMINI_API_KEY": "..."} (only required for the default Gemini mapper)

Available Tools

Tool	Description	Input
listResources	List all ingested resources.	{}
getStructure	Get the full JSON map of a resource.	{ resource_id }
getNode	Get metadata for a specific node.	{ resource_id, node_id }
resolve	Extract content (physical or virtual).	{ resource_id, node_id, virtual? }

---

See it in Action

Check out the Study Tool Example to see how an AI agent uses these tools to help you study a lecture video.

Troubleshooting

• "Command not found": Ensure citekit is installed in your PATH, or use absolute paths to your python executable.
• "Missing API Key": If you ingest with the default Gemini mapper, provide GEMINI_API_KEY in the env section of the config.