GitHub
Run your eval suite on every pull request and block the merge when scores drop.
Install the Fenwick app on the repositories that hold your prompts. Every pull request that touches a prompt file, a grader or a model setting starts a run against the datasets you pick.
The check posts the pass rate, the cases that changed and the cost difference as a comment. When a rule fails, the check turns red and the merge button waits.
Fenwick reads prompt files and writes check results. It never pushes code.
Category
Source control
Setup time
5 minutes
Plans
Every plan, Hobby included
More integrations
Run your next prompt change against 200 real cases this afternoon.
Free for 500 runs a month. Team from $39 a seat.