Building on the API
Fetch in bulk, not per row
Section titled “Fetch in bulk, not per row”For a page with many listings, fetch GET /v1/items/{def}/{paint}/patterns
once per item and look up each seed locally. Never make one /seeds/{seed}
call per row. For a full local copy, fetch
/v1/dump instead of crawling every
item.
Respect the cache
Section titled “Respect the cache”- Send
If-None-Matchand handle304. See Caching & conditional requests. - Responses are cached at the edge for 10 minutes. Polling faster re-reads the same response and spends your rate limit.
- Check
GET /v1/metaand re-fetch only whendataset_versionchanges.
Polling cadence
Section titled “Polling cadence”| you’re building… | check | cadence |
|---|---|---|
| a live overlay reacting to new guides | GET /v1/meta, compare dataset_version |
every few minutes, not every second |
| a nightly mirror or search index | GET /v1/dump with If-None-Match |
once a day |
| a single page load | whatever endpoint fits the page | once, on load |
Respect the rate limit
Section titled “Respect the rate limit”Spread requests out instead of firing them all at once. On a 429, wait
Retry-After seconds. In a browser, the edge burst limit arrives as a
network error: wait 10 seconds and retry. See
Access & rate limits.