Quick answer
Cursor provides a team admin API for member and usage metrics on business plans, and separate API keys for using models programmatically. The admin API reports aggregated usage, not per-prompt tokens. For individual plan usage, the dashboard or a local tracker is the practical route.
Two different APIs
| API | Answers | Scope |
|---|---|---|
| Team admin API | Member usage and spend | Business plans |
| Model API keys | Programmatic model calls | Per key |
| Dashboard | Plan and on-demand usage | Individual |
| Local tracker | Machine-level usage | Per device |
What you can build
- A weekly per-member usage export for finance.
- A spend alert that fires before the cap is reached.
- A dashboard that joins Cursor with your other tools.
- A chargeback report by team.
Endpoint names and scopes change, so read the current Cursor documentation before shipping anything that depends on them.
The gaps
There is no endpoint that returns per-prompt token counts. If you need token-level detail, you need a tool that records it locally. UsageAtlas covers the local side and shows Cursor's reported usage in the same window as your other providers.
For the manual equivalent, see the Cursor analytics guide.
What you can automate
Team plans expose admin metrics for members and spend. That is enough to build a usage export or a spend alert. It is not enough to reconstruct per-prompt tokens, because Cursor does not expose them.
- Automate a monthly per-member export.
- Alert before a spend cap is hit.
- Join Cursor data with other tools in one report.
- Do not promise token-level attribution the API cannot give.
Endpoint shapes change, so read the current documentation. The analytics guide covers the manual equivalent.
Questions, answered
Does Cursor have a public usage API?
Team plans expose admin metrics for members and usage. Individual plan usage is shown in the dashboard rather than a public endpoint.
Can I pull token counts from Cursor?
Not per request. Cursor reports usage in its own accounting units, and per-token detail is not exposed.
Is the model API billed separately?
API keys bill separately from your plan allowance. Check the current pricing for the models you enable.