What it does
Permission automation is off by default and is not a security boundary: missing or failed judgments let Cursor proceed and leave OpenCode's existing policy unchanged.
Agent tooling
jev-router
Suggest a relevant skill before a coding-agent turn, with optional Jev judgments on pending actions and completed work.
Bookmark: jev-router
Keep this in your collection.
Leave a noteWhat would you try with this?
: jev-router
Only you can see your notes.
Screenshot unavailable. Open the experiment ↗
How you can use it
For a coding assistant with many specialist instructions, borrow the idea of checking which instructions fit each new request. This tool first ranks the list, then checks a few choices and suggests one to the assistant. A developer would connect it to OpenCode or adapt its Cursor integration.
Those checks send recent conversation or pending commands to a paid outside AI service. Keep the ordinary permission checks in place: this tool may let work continue when an answer is missing or fails. Its suggestions do not force the assistant to follow them, and they do not prove an action is safe.