v1.2 → v1.3

This content is for v1.x. Switch to the latest version for up-to-date documentation.

v1.3.0 (2026-06-09) has no breaking changes. Embeds and workers upgrade drop-in.

What changed

The worker’s rate-limit response (429) now includes which window was hit:

{
  "error": "Too many requests. Please wait a minute before trying again.",
  "code": "RATE_LIMITED",
  "limitType": "minute"
}

with a Retry-After header (60 for the minute window, 3600 for the hour window). The widget uses this to show a window-specific error message and an accurate retry time.

Custom clients: if you call /api/chat directly, you can now branch on limitType — see the REST API reference. The field is additive; clients that ignore it keep working.