What it is
jev-guard is a security hook for coding agents including Claude Code, Codex, Copilot, Gemini, Cursor, pi, OpenCode, and ACP clients. It risk-scores tool calls, flags prompt injection in results, and checks skills and plugins.
How it uses Jev
Jev answers typed questions: risk (Score), approval (Noul), user_requested (Noul), and from_untrusted (Noul) over session context. The policy in src/guard.js denies, asks, or allows based on returned probabilities. Instruction files use INSTRUCTION_QUESTIONS.
Primitives:scorenoul
Technique worth stealing
Use typed Jev questions over session context to decide deny/ask/allow per tool call, with untrusted-content tracking.
Try it
npm i -g jev-guard; jev-guard key "…"; jev-guard install claude|codex|copilot|gemini|cursor|pi|opencode
Evidence
Each line is one question put to Jev about the README. ≥ 0.60 reads as yes, ≤ 0.40 as no; in between Jev is not making a call.
- Jev-centricyes0.88
- Shows a System One patternyes0.74
- Handles uncertaintyno0.08
- Measuredno0.05
- Runnableyes0.93
- Worth recommendingunclear0.54
- Model replicano0.07
- Problem scopescore on a 0–2 scale1.98
- About Jevyes0.98
Signals by Jev jev-1.13.0, card written by DeepSeek V4.1 Flash from the README on 20 Sept 2026.