Is It Us

When something breaks, find out whether it is upstream before assuming it is you.

check reads the public status pages for GitHub, Cloudflare and AWS, writes one record per provider, and rolls them up into a single summary. The rollup works out which open incidents are new since the previous check and carries a ready-made alert title, message and ntfy priority.

Alerts on what is new, not on what is open

A push fires when an incident appears that the last check had not seen. An outage that stays open for hours, or an AWS regional event that stays open for months, pushes once. A status page that is briefly unreachable keeps its incidents marked as seen, so it does not re-alert when it comes back. The first check alerts on everything already open, since that is when watching began.

Pieces

• @skunk-ape/is-it-us — the behavior. One read-only method, check: one GET per provider. It fails only when every provider is unreachable, since a rollup of nothing would read as all clear.
• @skunk-ape/is-it-us/rollup — a report attached to the behavior by default: one table of providers, their open incidents, new ones marked.

The push goes through @mgreten/ntfy-notify. The README has a workflow that runs check and then send, guarded so send is skipped when nothing is new.

Sources

GitHub and Cloudflare publish Atlassian Statuspage summary.json. AWS publishes its Health Dashboard feed as UTF-16 JSON, which is decoded before parsing. Scheduled maintenance is not counted as an incident.