用 Jev 与 CodeGraph 组合做安全审查
开发者分享了 typesafe-security-review 项目:把 Jev 的概率判断与代码图、OWASP/CWE 信息结合,用于低成本安全扫描。
针对 Jev-guard 的安全边界提问:仓库自带的 .mcp.json 或 hooks 是否可能在信任检查前运行,以及 Jev-guard 能否提前拦截。
Nice. One thing we'd test on it: a repo's own .mcp.json or hooks can run before any tool call reaches your guard. In headless mode Claude Code starts them before a trust check, we measured it on AgenticBench. Does Jev-guard sit early enough to catch that?