Slack error app_rate_limited: Events API deliveries rate limited

Your app exceeded the Events API delivery ceiling, and Slack is telling you so with an event. The documentation states: "Event deliveries currently max out at 30,000 per workspace/team per app per 60 minutes." Cross that line and, instead of your subscribed events, Slack sends a special app_rate_limited event to your Request URL, identifying the workspace (team_id), the app (api_app_id), and the minute in which limiting began (minute_rate_limited, a Unix timestamp rounded to the minute). This is not a Web API response — no request of yours failed; deliveries to you are being dropped.

That inversion changes the debugging posture completely. Nothing shows up in your outbound logs, because the loss is inbound. Symptoms are missing events: reactions unprocessed, messages unanswered, sync pipelines falling behind for exactly the noisy workspaces. If you subscribe to high-volume event types across a large workspace — every message in every channel, user_change, presence-adjacent floods — 30,000 per hour is closer than it sounds: that is 500 events a minute averaged.

Events dropped while you are rate limited are gone from the delivery stream; recovery means reconciling by polling the relevant Web API read methods for the affected window (mind their own rate limits). The durable fix is subscribing to less: request only event types the app actually consumes, filter early at your edge, and for very high-volume workloads consider whether each subscription is pulling its weight. Also verify your endpoint acknowledges within 3 seconds — a slow endpoint triggers Slack's retry machinery, and retried deliveries add to your volume, compounding the problem (see url_verification and endpoint health).

{
  "token": "Jhj5dZrVaK7ZwHHjRyZWjbDl",
  "type": "app_rate_limited",
  "team_id": "T123ABC456",
  "minute_rate_limited": 1518467820,
  "api_app_id": "A123ABC456"
}

Treat event subscriptions as a budget: subscribe narrowly, measure per-workspace event volume, and alert when any workspace trends toward the 30,000/hour ceiling instead of discovering it from losses. Keep acknowledgment sub-second by decoupling receipt from processing. The Events API numbers live in Slack Events API limits, and the silent-bot symptom chain in Slack bot not responding.