GitHub · by Openlayer

jevals

Agent evals and guardrails as one Jev request per trace.

Vishal Pandey

@_vMyth

Runs agent evals and guardrails as 1 request to a Jev decision model, not an LLM judge. Cheap enough to score every trace. Change 2 defaults: the quick-start sends your traces to a hosted API, and a gate lets calls through if it's down. github.com/openlayer-ai/j…

Oct 15, 2026 · 1 likesOpen on X

Openlayer’s library runs evals and guardrails for agents with a decision model in place of an LLM judge. All the checks for a trace go out as one request: tool choice, grounding, scope, relevance, completeness, indirect injection and PHI. In its quickstart, eight checks took one request of 1,388 tokens, cost $0.00006 and came back in 0.33 s. It works with Jev through TypeSafe, Vercel or OpenRouter, and locally with Kev or Laya.

Vishal Pandey flags two defaults to change: the quick-start sends your traces to a hosted API, and a gate lets calls through if it is down.