Cursor setup

Open Cursor Models, then enter:

• Base URL: https://api.strixgate.dev/v1
• API key: your Strix API key
• First model: strix-auto

If the link does not open, launch Cursor and go to Settings → Models. Find the OpenAI API key section, enable Override OpenAI Base URL, paste the values above, and add each custom model ID manually.

Cursor sends OpenAI-compatible requests to /v1/chat/completions. Add the IDs you want visible in the model picker. A clean starter set is:

• strix-auto — automatic routing for mixed work
• strix-quick — Fast mode for small, focused, reversible tasks
• strix-coding — everyday coding and tool-use loops
• strix-deep — complex multi-file work
• strix-bug-hunt — debugging and root-cause analysis

Recommended selection

Use strix-auto for mixed work. Choose strix-quick when you want fast responses for simple tasks. Select another explicit model when the intent is clear—for example, debugging with strix-bug-hunt or UI work with strix-visual.

The available authenticated catalog is returned from GET /v1/models. Routing evidence appears in x-strix-* response headers; the provider choice itself is intentionally abstracted behind the lane.