Submit a project
jev-sentinel
By harshwasan
Source description: Pi coding-agent extension: TypeSafe Jev checks for tool calls, tool outputs and replies (prompt injection, approvals, secret scrubbing, task pinning)
Repository description
The repository describes its purpose as: Pi coding-agent extension: TypeSafe Jev checks for tool calls, tool outputs and replies (prompt injection, approvals, secret scrubbing, task pinning)
Source evidence
The daily check found an explicit Jev / TypeSafe AI connection in the original README and implementation files at the same commit.
A guard for coding agents, built on TypeSafe's [Jev](https://docs.typesafe.ai/) model: it checks what the agent does, reads, and says. One set of checks, three hosts: a [Pi](https://github.com/earendil-works/pi) extension, and a plugin for **Claude Code** and **Codex CLI**.
Automatically source-checked: public repository, pinned README, and implementation files. Not run, benchmarked or security-audited.
