Measured, not marketed.

Not measured since 2026-09-15 04:46Z

Cadence 15 min. A finding open is a control reporting red, not an outage; the register is internal and read by the operator.

Last successful ping 2026-09-15 02:04Z; 1637 of 4321 succeeded over the last 30 days (37.9%)

Last attempt 2026-09-15 05:44Z. Hourly. The ping succeeds only when every required scheduled job reported an outcome inside its window — a single job missing its window fails the ping for the hour.

No 90-day record exists yet; the window above is the whole record held.

3 of 3 active producers landing rows of their declared kind inside their declared expectation

Hourly dead-man check across all active producers, counted as one aggregate. Producers are not named here.

Last completed run 2026-09-10 (full), finished 2026-09-11 02:28Z (resumed by catchup)

Manifest present. Artifact monitor reads that run as "run not reconciled" at 2026-09-15 05:07Z. 4 scheduled run(s) since have not completed.

Last run independently read back and hash-verified: 2026-08-28 (receipt N-724).

Nightly at 07:15 UTC with a 20-minute catch-up. Scope: five-year rollup record, configuration and tenancy — not the 72-hour raw window.

https://cyntech.io serving commit 9a3612828f2c; deployment scan verdict "failed", 1 critical finding(s)

Observed every 20 minutes by the publish watch. This is the platform security scan taken against the deployed build, not the CI run on main — the CI conclusion is not yet stored in a record this page may read, and is therefore not shown.

Critical finding class: integration_error_state. The class is published so the count is not read as something it is not; "integration_error_state" is a tenant integration in an error state, not a defect in the deployed build. The verdict stays failed until it is repaired.