- Category
- Developer Tools
- Rank
- No. 3349Tools index
Previous survey · No. 3180 ·
- Pricing
- Open Source
- Type
- TOOL
- Use case
- Model & Agent Evaluation
- GitHub
- 111 stars
- Date
About
jevals runs evals and guardrails for a full agent trace in a single request, using Jev-style decision models instead of an LLM-as-judge. It bundles agent checks such as tool choice and groundedness, quality checks such as relevancy and completeness, and security checks such as indirect prompt injection and PHI exposure, with each request costing a few thousandths of a cent and returning in a few hundred milliseconds, so it can run on every trace inside the agent loop.
Why it made the leaderboard
It lets teams run comprehensive agent evals on every trace in real time rather than only after the fact, since each check costs a fraction of a cent.
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.