Skip to Content
TutorialsLoad first dataset with Coding Copilot and MCP

Load Data with Coding Copilot

We assume you set up your cognee mcp server, if you didn’t got back here.

Let’s open Cursor

Below is one possible setup that works well on macOS. Feel free to adapt the paths to your own file system.

1. Register a new Cognee MCP server

Create a helper script named run-cognee.sh at your prefered location to start the server:

#!/bin/bash export LLM_API_KEY="YOUR_OPENAI_API_KEY" uv --directory /Users/<YOUR_COGNEE_PATH>/cognee-mcp run cognee

Now open the MCP settings, navigate to Tools & Integration and click on New MCP Server. Add the cognee in the mcp.json:

{ "mcpServers": { "cognee": { "command": "sh", "args": [ "/Users/<YOUR-HELPER-SCRIPT>/run-cognee.sh" ] } } }

Cursor will automatically start the new instance for you.

2. Load the Rust documentation

  1. Download the Rust documentation (or any Rust article) and save it as rust.md inside your project.
  2. In the Cursor chat run:
cognify rust.md

Approve the tool execution when prompted. You can monitor the ingestion progress with:

cognify_status

Wait until cognify is completed.

3. Ask Agent to help

With the content now in memory write the best function based on rust

The agent will analyse the freshly ingested documentation and suggest an optimal Rust implementation.


Next steps