Set up Cursor with PentestLLM

Cursor’s help page on own API keys, read on 11 November 2026, documents adding a key per provider under Cursor Settings, Models. It does not document the base URL field. The label Override OpenAI Base URL and the custom model step come from threads on Cursor’s own forum from August and September 2026. If your version of Cursor has no such field, use Cline inside the editor instead; it treats a custom endpoint as a normal provider.

The override is global

A Cursor forum bug report from August 2026 says that while the override is on, requests for Cursor’s own OpenAI-family models also go to your endpoint and fail there. Turn the override off when you switch back to Cursor’s models.

Chat models only

Cursor’s help page says custom API keys only work with chat models and that Tab completion keeps using Cursor’s built-in models.

Requests pass through Cursor

The same page says your key is sent to Cursor’s backend with every request, because requests are routed through its servers for prompt building, and is not stored there.

Plan limits

Cursor’s pricing page says its Plus plan does not include your own API keys, and that Teams and Enterprise plans add Cursor’s own token rate to own-key requests.

Privacy mode

Cursor’s help page says its Zero Data Retention policy does not apply when you use your own API keys.

If you want the router and Cursor will not take auto/security as a model name, create a key in the console with its auto router option set to auto/security. A key set that way routes every chat request as the router id, whatever model name the client sends, so the named build in the steps becomes a label and the router still picks per request.

Requests bill per token to your PentestLLM balance at the price of the build that answered; the table below is the live catalog on 2026-11-11. Agent requests carry the conversation, the files Cursor read and tool results, so input tokens are most of the bill.

The builds answer security questions other models decline, which is what makes them useful for reviewing code for vulnerabilities and drafting findings inside the editor. Refusals are reduced, not removed. Use them on code and systems you own or have written permission to test, as the acceptable use policy says. The other agent guides cover terminal tools.