Home /Safety, moderation and verification /jev-guard

jev-guard

Auto mode for Claude Code, Codex, Cursor, Gemini CLI, Pi, and OpenCode: risk-scores each tool call as deny, ask, or allow and flags prompt injection in results.

Open on GitHub Entry 337 in the readme

Details

Section: Safety, moderation and verification
owner: leepokai
stars: 12
forks: 1
pushed: 2026-09-17
Language: JavaScript
License: MIT

What it is

Form: Hook or plugin
Host agent: Any agent
Audience: End users

Maturity

Docs

reports measured numbers

Useful for

Tool gating and guardrails0.91

Moderation and safety0.83

Output verification0.81

Best intent matches

gate or approve tool calls before they run0.92

detect prompt injection or risky commands0.89

judge or verify an agent's output0.83

moderate content or detect abuse0.79

drive a browser or GUI with an agent0.69

+ Submit a project ★ Star on GitHub