Uptime, upstream status and outage timeline (root key only)
GET /admin/health-history · Admin API key required
Reads the health probe history. A cron runs the same checks as `GET /health` every 5 minutes and stores one row per run. Returns uptime percent, coverage (how many of the expected probes actually ran), per-upstream uptime and probe latency (average, p95, max), an outage timeline (runs of consecutive degraded probes, newest first) and daily uptime. `current.stale` is true when the latest probe is older than 15 minutes, which usually means the cron is not running. A probe that could not reach the database is held in KV and written with the next one, so a database outage still shows up. Only Supabase, Cohere and Groq count towards an outage. Needs the root key. When storage is not ready the response has `available` false and a note.
Parameters
hours(query) - Look-back window in hours.
Responses
200- Health history summary401403- Not the root key