Cursor IDE Integration

Cursor allows custom models and custom base URLs, making it a first-class citizen for Muxa integration.

Configuration Steps

1. Launch Muxa locally (or remotely via an internal network). Example: http://localhost:8081.
2. Open Cursor Settings.
3. In the settings, navigate to Features -> Models.
4. Set the Base URL to http://localhost:8081/v1 and the API key to sk-muxa.
5. Toggle off all default models and explicitly add the model name configured in your Muxa deployment (e.g., gpt-4o, claude-3-5-sonnet-20241022).

Enabling Codebase Indexing (Embeddings)

To use the @Codebase feature effectively with Muxa, you must configure Muxa to act as an embeddings provider using a fast, cheap model (like local Ollama nomic-embed-text or OpenAI text-embedding-3-small).

See the Embeddings Configuration guide for setup instructions.

Once Muxa is serving embeddings, Cursor will automatically route its massive background indexing jobs through Muxa, where you can easily monitor and cache the traffic.