Jev in production › Agent guardrails and approvals

Jev LLM Guard

Jev scores user, tool and model text against OWASP LLM Top 10 risks, returning allow, review or block, with a live Turkish and English demo.

Open on GitHub ↗

The source publishes no measured number.
Use
Agent guardrails and approvals
Industry
Security
Form
Open-source tool
Stage
Beta
Listed
2026-10-01
Found via
github
Repository
gulbaki/jev-llm-guard
Stars
14
Forks
1
Last push
2026-10-01
Language
JavaScript
License
MIT

The README opens with

A contextual LLM guardrail powered by Jev System One. Submit text as user input, retrieved content, tool output or model output and receive a local allow / review / block decision plus ten OWASP risk assessments.

Try the live demo — enter Turkish or English text, select its context, and inspect the decision and JSON result.

The taxonomy follows OWASP LLM Top 10 2026. Scores represent textual risk signals, not calibrated vulnerability probabilities. Application-level risks that cannot be verified from the supplied information return needscontext. Jev provides semantic scores; this package validates the response and applies a versioned decision policy.

Badge

For the project's own README, linking back here:

Listed in Jev in production

Also used for agent guardrails and approvals