# jevals

Runs a trace's agent, quality, and security evals as one batch of typed Jev questions instead of separate LLM-judge calls.

- URL: https://github.com/openlayer-ai/jevals
- Section: Benchmarks, evals and calibration
- Type: project
- Stars: 111
- Pushed: 2026-11-02T23:37:32Z
- Language: Python
- License: MIT
- First seen: 2026-09-27
- Form: Library or SDK
- Useful for: Output verification (0.94), Benchmarks and evals (0.93), Tool gating and guardrails (0.91)

Page: https://awesomejev.vercel.app/p/jevals/