Public API

Paging through results and staying inside the limits

Listing endpoints page with a cursor rather than an offset, so follow the cursor the response gives you instead of computing one. Watch the rate-limit headers, and use updatedSince for polling rather than re-reading everything.

Reviewed August 20, 2026

Steps

1. Follow the cursor from each response until none is returned. Do not construct cursors yourself.
2. Keep page size reasonable — larger pages mean fewer requests but slower ones.
3. Poll with updatedSince rather than walking the whole collection on a schedule.
4. Read the rate-limit headers and slow down before you are refused, not after.
5. Store the requestId of anything that fails, for support.

How to check it worked

A full walk of a collection returns each object once, and repeated polling reads only what changed.

What usually goes wrong

• Building offset pages, which drift as data changes underneath.
• Re-reading everything every minute and hitting the limit.
• Reusing a stale cursor and getting an invalid-cursor error.