rate

## Automatic Jev check When `/rate` is reviewing an actual branch diff and has an evidence-backed preliminary rating for a named guideline, read `${AGENTIC_HOME:-$HOME/.agentic}/artifacts/jev/PROTOCOL.md` and run its helper. Send stable source IDs, the bounded diff/context, the named guideline, and the proposed rating. Ask a **Choice**: `Does the supplied evidence support this proposed rating?` with `supports` (the cited evidence supports it), `contradicts` (the cited evidence materially conflicts), and `unclear` (context is insufficient). Reconcile the advisory answer with the evidence; retain the original rating and explanation on unavailable, contradiction, or unclear. When `/clean`, `/ddd`, or `/oop` are borrowed criteria, do not invoke their separate Jev checks; still run this `/rate` check. It never changes the visible 1–5 scale or authorizes changes. Before scoring, say that you are using `/clean`, `/ddd`, and `/oop` as review criteria. These are borrowed criteria, not child-skill invocations. According to /branch, rate from 1 to 5 (5 means perfect, 1 means terrible) how much my code follows principles for each principle shown in: ## Bob Martin guidelines /bob ## Clean code guidelines /clean ## Domain driven design guidelines /ddd ## Object oriented guidelines /oop ## Idiomatic guidelines /idiomatic Also, make sure to explain each of the terminologies you are using.