What problem does it solve? When a multi-agent team is kept alive by scheduled reconciles spread across several machines, operators cannot tell whether anything is actually running or which host silently went dark. This skill answers those questions deterministically with preflight checks, fleet health folds, operator digests, and role-vacancy escalation. ## Core Features & Use Cases - Doctor preflight: coord-engine doctor verifies the storage CLI is on PATH, the File Store is reachable, and reports the engine version — run it after install and inside scheduled job self-tests. - Fleet health fold: coord-engine health aggregates per-host health shards (last-reconcile age, STALE flag over 24h, engine version) and exits non-zero when no host is fresh, making it usable as a monitoring probe. - Operator digest: coord-engine digest folds blocked-on-you items, upcoming tasks, agent liveness, and stale tasks into a report, optionally persisted to _coord/digests/ or emitted to a timeline track with idempotent, heartbeat-safe retries. - Role-vacancy escalation: coord-engine escalate sweeps role docs and files a P1 directive to a vacant role's maintainer once its SLA lapses, idempotent per day. - Use Case: Wire health --json into a heartbeat wrapper and alert on healthy: false to catch a reconciler that stopped on one machine before the shared index goes stale. ## Quick Start Run the doctor preflight and fleet health check for my fulcra agent team, then generate today's operator digest showing what is blocked on me.
