Meta

Reports whether the database answers and whether each background worker has finished a loop recently. A worker is ok when its last successful iteration is within three of its loop intervals, and never sooner than two minutes, so that a queue taking a minute to drain is not called an outage. status is degraded when any check is not ok. No API key needed. The answer is cached for 10 seconds, so time can be up to 10 seconds behind. Both health endpoints share one limit of 60 requests a minute per IP.

The health of the API is sampled once a minute and kept for 90 days. This endpoint aggregates those samples per UTC day. What is measured is minutes, not samples: a day owes 1440 of them, today owes the minutes already behind it, and a minute with no sample counts as degraded exactly like a sample that came back degraded. A day owes minutes only from the first sample ever taken onwards, so a day that was over before sampling began is unmeasured and comes back with uptimePercent null. Every day of the window is in the answer, so a day that was sampled and came back with nothing is 0 percent rather than missing. No API key needed, and the answer is cached for 60 seconds. Both health endpoints share one limit of 60 requests a minute per IP.

Public numbers about yeetpost as a whole, which is what the marketing site's proof pill draws. Nothing here is about one account: postsThisWeek and postsAllTime count posts that reached a platform (no error, not a draft, not taken down, not still waiting in a queue), sandbox connections and test-key traffic left out, and liveConnections counts the connections that can be posted to right now. The counts are whole numbers exactly as counted, so round them yourself. No API key needed. The answer is cached for 60 seconds, and this endpoint has its own limit of 60 requests a minute per IP.

Returns this OpenAPI document. No API key needed.