joelclaw System Health Check

Run scripts/postboot.sh for the read-only postboot/daily topology sweep. It checks the Flagg/Panda authority split, NAS mounts, NFS tuning, custom NAS MinIO, Central Postgres, local Convex, Typesense, Inngest, and shadow Central health without printing secrets. The daily automation also reviews recent Codex infra transcripts so completed work does not stay trapped in chat.

Migration caveat — 2026-07-09

The current health script is still too Panda-centric: it treats Panda-era k8s/PDS/NodePort surfaces as if they were the whole joelclaw system. That is no longer aligned with the Central migration vision.

Health needs to become lane-based:

• Central lane: Flagg/Mac Studio native launchd services, Redis, Typesense, Inngest, system-bus worker, agent-mail, run capture, session NAS backup receipts, reboot-survivable state.
• Relay lane: Panda-only leftovers such as iMessage/account-bound relays, legacy webhook/callback surfaces, and explicit decommission blockers.
• Satellite lane: Blaine/other Machines' JOELCLAW_CENTRAL_URL, outbox backlog, fresh capture timestamps, and transcript backup freshness.

Until that rewrite lands, read red Panda/k8s/PDS checks as migration TODOs, not proof that Flagg Central is dead.

~/Code/joelhooks/joelclaw-runtime/skills/joelclaw-system-check/scripts/postboot.sh