pi-jev-sentinel
Pi의 도구 호출, 도구 출력, 응답을 Jev로 검사하여 위험한 작업과 프롬프트 인젝션을 잡아내며, 사용자 승인, 컨텍스트 재검사, 시크릿 제거, 선택적 작업 고정을 제공합니다.
이 페이지는 영문 목록에서 생성했습니다.
readme에서
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 …
상세
- 섹션
- 코딩 에이전트
- owner
- harshwasan
- 스타
- 15
- 포크
- 0
- 최근 커밋
- 2026-09-20
- 언어
- TypeScript
- 라이선스
- MIT
어떤 프로젝트인가
- 형태
- 훅 또는 플러그인
- 호스트 에이전트
- Pi
- 대상
- 에이전트 개발자
성숙도
문서