What it does
A score of 0.7 or more adds a delegation hint. If Jev is unavailable, the hook returns to its regex-based score.
Agent tooling
CC Settings asks Jev whether a Claude Code prompt is broad enough to delegate.
Only you can see your notes.
Screenshot unavailable. Open the experiment ↗
A score of 0.7 or more adds a delegation hint. If Jev is unavailable, the hook returns to its regex-based score.
Your developer can adapt this project to check if a user instruction is broad. The code sends the prompt to an outside AI service called Jev. Jev evaluates the text and returns a simple number score. If this score is high enough, the tool suggests assigning the work elsewhere.
This setup requires an access key to connect the tool to Jev. If the connection fails or takes too long, the code skips the AI service entirely. It relies on a basic word search to score the prompt instead.