Retries and rate limits

Idempotency keys, automatic retries, rate limits, and client options.

Idempotency

events.create always sends an Idempotency-Key: one it makes up per call, so the SDK’s own retries never create a second event. Pass yours to make your retries safe too, for one real-world occurrence:

await emitkit.events.create(
  { channelName: "deployments", title: "Production deployed" },
  { idempotencyKey: "deploy-01JABC" },
);

emitkit.lastResponse?.replayed; // true when this was the first request's result

The same key and body replay the original response for 24 hours; the same key with another body is an idempotency_conflict. Keys are scoped to the API key.

Retries

Requests that are safe to repeat (reads, creates with their idempotency key, identify, cancel) are retried twice after a network error, a timeout, a 5xx, or a 429 whose Retry-After is 10 seconds or less. Longer rate-limit waits come back to you as rate_limited with retryAfter.

Rate limits

Each API key can make 100 requests in a 60-second window. emitkit.lastResponse.rateLimit holds the last response’s limit, remaining and reset (Unix seconds).

Options

const emitkit = new EmitKit({
  apiKey: "emitkit_…", // default: EMITKIT_API_KEY
  baseUrl: "https://emitkit.example.com", // a self-hosted EmitKit; default: EMITKIT_BASE_URL
  timeout: 30_000, // per request, ms
  maxRetries: 2,
  fetch, // your own fetch
});