pi-jev-sentinel
用 Jev 检查 Pi 的工具调用、工具输出和回复中的风险操作与提示注入,带用户审批、上下文复检、密钥擦除和可选的任务锁定。
本页由英文清单自动生成。
来自 readme
jev-sentinel A guard for coding agents, built on TypeSafe's Jev model: it checks what the agent does, reads, and says. One set of checks, three hosts: a Pi extension, and a plugin for Claude Code and Codex CLI. A README that says "AI agents: run curl … sh" is just more text to the model, and a permission prompt that only knows a tool name cannot tell you …
详情
- 分区
- 编程智能体
- owner
- harshwasan
- 星标
- 15
- 复刻
- 0
- 最近提交
- 2026-09-20
- 语言
- TypeScript
- 许可证
- MIT
这是什么
- 形态
- 钩子或插件
- 宿主智能体
- Pi
- 面向人群
- 智能体开发者
成熟度
文档