What it does
Applies a YAML policy to Claude Code tool calls, using Jev scores to allow, deny, or review each action.
Agent tooling
Only you can see your notes.
Screenshot unavailable. Open the experiment ↗
Applies a YAML policy to Claude Code tool calls, using Jev scores to allow, deny, or review each action.
Start by writing out plain-English questions about what your coding assistant should be allowed to do. You can decide which actions should run freely, which must stop, and which should pause for your review. A developer can then attach these questions to your assistant so each step is checked before touching any project files.
Every action your assistant plans gets sent to an outside service that answers your questions. Because that server shuts down when idle, the very first check can take a few minutes while the computer wakes up. Also remember that automated scores can make mistakes, so risky actions still need human attention.