Cursor MCP integration

The user-facing part.

Bring your video corpus into Cursor. Register the helper as an MCP server to reference transcripts and documentation directly in your editor. You can prompt Cursor to write code matching a video walkthrough, citing sources as you write. Everything runs on your machine, protecting your workspace.

Run the loop.

1. Open Settings in Cursor.
2. Navigate to features and select MCP.
3. Click Add New MCP Server.
4. Paste the Uoink command, args, and env from the stdio config, using absolute paths.
5. Set UOINK_MCP_TOOLSET=core in the server env to use the previous 35-tool set under Cursor's reported 40-tool cap.
6. Ask Cursor's chat panel to search your library.

How Uoink handles it.

Cursor runs the bundled Python against uoink_mcp.py as a stdio subprocess. Tools such as search_library and get_library_item let the client search and read saved source material.

Agent-readable surface.

{
  "name": "search_uoinks",
  "arguments": {
    "query": "Model Context Protocol"
  }
}

Full server metadata lives at /mcp and /mcp/manifest.json.

Run it on your next source.

Configure Cursor to query your transcript corpus directly from the editor.