Building an AI agent? Start with /llms.txt for the full site index.
Docs
Request rates, quota control, and how to integrate under high concurrency.
A rate limit governs how many requests you may make per unit of time. A quota governs the total available across a plan period.
Both apply at once. Traffic that looks fine minute to minute can still exhaust the period total.
The TWSE and TPEx daily price endpoints currently allow 60 requests per minute per API key.
High-frequency queries, historical backfills, and several services sharing one key are the three situations that hit a limit first.
Implement a concurrency pool and request throttling on the client side.
Handle 429s and retry intervals once, in the SDK or at the API gateway, rather than reimplementing the logic in every service.
Local caching, batch aggregation and incremental queries cut request volume substantially.
Do not retry blindly at a fixed interval — that usually amplifies congestion rather than clearing it.
Set an alert threshold in the dashboard so load can be shed before the quota runs out.
