Skip to content
System One

leepokai/jev-guard

Security hook for coding agents that risk-scores every tool call with session context (deny/ask/allow), flags prompt injection in tool results and checks skills and plugins. Supports Claude Code, Codex, Copilot, Gemini, Cursor, pi, OpenCode and ACP.

78-second launch video: every tool call risk-scored with session context, prompt injection flagged, skills checked
Image from github.com

leepokai/jev-guard on GitHub

Stars
30
Forks
4
Language
JavaScript
License
MIT
Last push
17 Sep 2026

Read from GitHub on . Counts change daily.

How it uses Jev

// One fetch, two backends: TypeSafe's API when JEV_API_KEY is set, Vercel AI Gateway otherwise.
// Question ids are ours; types are TypeSafe's (noul / choice / score). Gateway calls noul "boolean".
const TYPESAFE_URL = "https://api.typesafe.ai/v1/systemone";
const GATEWAY_URL = "https://ai-gateway.vercel.sh/v4/ai/evaluation-model";

import { readFileSync } from "node:fs";
import { homedir } from "node:os";
import { join } from "node:path";

export const CONFIG_FILE = join(homedir(), ".jev-guard", "config.json");

// Env first (CLIs inherit the shell); then ~/.jev-guard/config.json written by `jev-guard key`, which is what
// GUI hosts such as Cursor or Zed need since they don't see your shell profile.

View in src/jev.js

Open on GitHub

More like this

Use cases this is tagged with