Jev in production › Agent guardrails and approvals

Canny

Hooks Claude Code and Codex CLI; Jev answers whether an edit breaks a CLAUDE.md or AGENTS.md rule and whether a message claims done, as notes only, while ledger checks block. A 25-pair Opus 5 A/B had 25 of 25 passes in both arms (author).

Open on GitHub ↗

25 of 25measured against a baseline, as published by the source
Use
Agent guardrails and approvals
Industry
Developer tools
Form
Open-source tool
Stage
Beta
Plugs into
Claude Code, Codex CLI
Listed
2026-09-21
Found via
github
Repository
qkal/Canny
Stars
122
Forks
14
Last push
2026-09-22
Language
TypeScript
License
MIT

The README opens with

A supervision layer for AI coding agents. It hooks into Claude Code and Codex CLI, keeps a ledger of what the agent actually did, and will not let it finish on a claim.

That is Claude Code, verbatim, during this project's first live run. It had been asked to add a function, it wrote the file with a shell heredoc, ran nothing, and finished. No error. No warning. Nothing in CLAUDE.md could have stopped it, because a rules file only asks the model to remember, and nothing checks that it did.

Badge

For the project's own README, linking back here:

Listed in Jev in production

Also used for agent guardrails and approvals