| name | diagnose-collapse |
| description | Collapse overview for a failing Clowder soak — names the cluster pattern (e.g. "189-style schedule-edge"), surfaces the collapse trajectory, and suggests next-step drills. Use when the user asks "why did the colony collapse?", "what kind of failure is this?", or "does this look like a prior cluster?". Do NOT fire for — pass/concern/fail gating (that's `just verdict`), per-tick drilling (that's `just q`/`/logq`), or cross-run SQL (that's `just logdb-query`). |
Collapse Diagnosis (/diagnose-collapse)
/diagnose-collapse provides a structured overview of a colony collapse. It reads a failing run's death / anomaly / action signals and produces a cluster-labelled narrative — naming whether the collapse resembles a known prior cluster and suggesting drills to confirm.
When to fire
Fire when:
- A soak failed (
just verdict returned fail) and the user wants the full collapse overview rather than a single targeted drill.
- User asks "what kind of collapse is this?", "does this look like the 189 cluster?", "why did the colony die?"
- Used as the balance investigation entrypoint when the failing pattern is not yet named.
Do NOT fire when:
- User wants the pass/concern/fail gate — that's
just verdict.
- User wants to drill into a specific tick, cat, or event — that's
just q / /logq.
- User wants cross-run aggregate patterns — that's
just logdb-query.
Relationship to neighbouring tools
just verdict — pass/concern/fail gate. Run verdict first; fire /diagnose-collapse when verdict returns fail and the next_steps suggest a full overview.just q / /logq — per-tick drill-down. /diagnose-collapse names the cluster; just q confirms the tick-level evidence.just similar — semantic retrieval across prose artifacts. After /diagnose-collapse names a cluster (e.g. "189-style schedule-edge"), use just similar <prior-ticket-id> to surface the prior fix, balance threads, and related tickets — e.g. just similar 189 --corpus landed,balance. This is the canonical "does this collapse pattern look like a prior cluster?" lookup.just logdb-query — cross-run SQL. /diagnose-collapse diagnoses one run; logdb compares patterns across many runs.
