Independent outside-in availability probe for hserver public services.
The workflow runs on GitHub-hosted infrastructure every five minutes, offset from the top of the hour. It accepts HTTP 200, standard redirects, 401 and 403 as proof that the public edge is alive, matching the production Blackbox policy.
Transport failures, TLS failures, timeouts, 5xx and unexpected status codes fail the workflow. The workflow opens or updates an assigned GitHub incident issue while the outage remains active, then comments and closes that issue automatically when all probes recover.
No hserver credentials, internal configuration or alert-delivery secrets are stored in this repository. A manual force_failure input exists only to test the incident lifecycle.
Scheduling note: GitHub scheduled workflows are best-effort; GitHub documents that scheduled runs can be delayed or dropped during high load. This watcher is therefore an outside-in signal, not a hard 5-minute SLA.