Reliability · in the open

Status

A status page for a person. Every feed on this site carries a freshness contract; the fleet behind it re‑verifies its own guards daily. State below is computed live, in your browser, from the same JSON the site itself reads — if something rots, this page shows it red.

Site data feeds

Contract: refreshed daily by a scheduled engine; age beyond 2× the interval (48 h) is an incident — per the same data‑integrity rule the fleet runs on.

Fleet discipline

Deterministic gates on the machine that builds all of this — re-run and republished daily as aggregates (counts only; details never leave the host).

How this works. Detection is deterministic and cheap — generated_at freshness contracts, an RCA mechanism health gate, dirty‑tree audits, launchd liveness checks. Agents are reserved for diagnosis when a sensor trips, never for detection: a watcher that can fail silently is the exact failure class this fleet exists to kill. Every incident that survives to a root cause becomes a runnable mechanism test, and those tests are what the "RCA mechanisms" count re-verifies daily.

Related: the trust audit — this site inspecting itself in your browser. ← back