Checks each tool call a coding agent is about to run, and only the unclear ones go further.
@chinesepoweredFlinch looks at every tool call a coding agent wants to make before it runs. Jev answers a set of risk questions in one request. Only the unclear calls are passed to a slower model.
It is wired for Claude Code today.
It was built at the San Francisco Jev hackathon.
A Claude Code hook that asks Jev about every shell command and blocks the ones that would delete something that matters.
Jev
Python
Claude CodeStops a coding agent from finishing on a claim: it hooks into Claude Code and Codex, keeps a ledger of what the agent really did, and asks Jev only for the judgment calls.
Jev
TypeScript
Node.jsClaude Code
CodexScores each action an agent wants to take against the job it was given: allow, ask a person, or block.
JevPythonClaude CodeCodexA demo gate for an agent that is about to act. With no Jev key, it uses a stand-in judge and still runs.
JevTypeScriptShows a captured page crash and asks Jev who owns it, whether a contract broke, and how far it spreads.
JevTypeScript
ReactA loop where a coding agent writes, a review comes back, and Jev decides if it is safe to merge.
JevPython
CodeRabbit participant
