Sign up
Docs menu

Cursor

Wire: OpenAI. Base URL: https://sator-api.princep.org/v1. State: reported, not verified by us. Cursor has the setting, but does not document it; what follows is how users report it working.

Install#

cursor.com.

Configure#

In Cursor: Settings → Models.

  1. Under the OpenAI API key section, enter your Sator key as the OpenAI API key.
  2. Turn on Override OpenAI Base URL and enter https://sator-api.princep.org/v1.
  3. Add a custom model with a Sator id from Models, for example deepseek-v4-flash, and enable it.

Verify#

Open a chat, select the model you added, and send Reply with exactly: ok. A working setup answers ok, and the request appears in your dashboard.

What to know#

  • Requests go through Cursor's servers. Cursor calls Sator from its own backend, not from your machine; the key you enter is held by Cursor. Use a key made for Cursor alone so it can be revoked on its own.
  • Chat models only. The override applies to chat and composer requests. Cursor's tab completion and some other features use its own models regardless.
  • Unverified. If the setting has moved or behaves differently in your version, that is the tool's UI changing, and support@princep.org would like to know.

Troubleshooting#

  • Every request fails — the base URL is missing its /v1, or the override toggle is off.
  • The model is "not available" — the custom model name must match the catalog id exactly. Check GET /v1/models.