agg-statuslisted

# /agg:status — report the AgenticGoGo loop status Give the user a concise, accurate status of the running (or last) `agg run` loop. ## Step 1 — Read the live state The loop writes a compact snapshot to `agg/state/state.json` in the project. Read it: ```bash cat agg/state/state.json 2>/dev/null ``` If it's missing, the loop hasn't been started — tell the user to run `/agg:new` (to set up) or `agg run` (if already configured), and stop here. You can also run the dry-run scoreboard directly (re-evaluates every judge right now): ```bash agg plan ``` Use `agg plan` when the user wants a FRESH evaluation; use `agg/state/state.json` when they want the state of the currently-running loop without re-judging. ## Step 2 — Report Summarize from the state JSON: - **Headline**: `N/M judges met` and the `done_if` condition (are we close?). `done_if` is the project's Definition of Done, composed from judge names. - **Per judge**: name, current measure (`18/28`, `82%`, `yes`), lifecycle state (met ✔ / in_progress ◑ / regressed ⚠ / pending ·), and any `▲+N` delta. **Call out regressions loudly** — a judge that was met and broke is the most important signal (the gate rolls that session back). - **Current step**: which step is running, and its **agent + model** (a mixed sequence runs different agents at different steps — say which one is live). - **Run health**: current `session`, `phase`, `idle_secs` (flag if ≥240 = possible stall), `tokens_spent` vs the `sequence.limits.t