## TL;DR

The org exhausted its 24-hour API request allowance, so every API call fails until the window resets. Check the limits endpoint to confirm, identify which integration burned the quota, then cut its call volume or spread the load. This is an org-wide budget problem, not a code bug.

## Error

```text
[ {
  "message" : "TotalRequests Limit exceeded.",
  "errorCode" : "REQUEST_LIMIT_EXCEEDED"
} ]
```

## Steps

1. Call GET /services/data/vXX.X/limits and read the DailyApiRequests Max and Remaining. Expected: Remaining near zero confirms the diagnosis.
2. Check which integration consumed the quota: API usage reports in Setup show calls by client and user. Expected: one or two consumers dominate, often a new agent loop.
3. Stop or throttle the heavy consumer immediately. Expected: no new calls burn while you plan.
4. Reduce call volume structurally: batch with the Bulk API, use the composite API to combine calls, and cache describe results. Expected: the same work in a fraction of the calls.
5. Schedule heavy loads for off-hours and stagger them so they do not overlap. Expected: the daily budget covers the whole day.

## When to use

- All API calls fail with REQUEST_LIMIT_EXCEEDED at once.
- A new agent or sync started recently and quota vanished faster than usual.
- End-of-day loads fail while morning loads succeed.

## When not to use

- "Too many SOQL queries: 101" (per-transaction governor limit).
- UNABLE_TO_LOCK_ROW (contention, not quota).
- A single integration failing while others work (that is not the org limit).

## Tool compatibility

- Salesforce REST, SOAP, and Bulk APIs; the limits resource.
- Any integration client: ETL, agents, Data Loader.

## Variant phrasings

### TotalRequests Limit exceeded

The long form; check the limits endpoint for the reset.

### API limit exceeded org-wide

Same condition described from the admin side.

## Why it happens

API limits are per rolling 24 hours per org. Agents that poll, chatty sync loops, and overlapping bulk jobs can burn a day's budget in hours, and then everything API-based stops.

## Edge cases

- The limit resets on a rolling basis, not at midnight; check the limits endpoint for the real recovery time.
- Sandbox limits are much lower; a load sized for production fails in a full sandbox.
- Buying more API calls is an option, but fixing the chatty consumer is cheaper and permanent.

## Provenance

Resolved from the public thread: https://vectle.com/posts/pst__CDPUQUxed5301IlZSaSpw