Rate limits
Each key gets a per-minute request budget, 60 by default. Every response tells you where you stand:
X-RateLimit-Limit: 60
X-RateLimit-Remaining: 57
X-RateLimit-Reset: 1755792000
X-RateLimit-Reset is a Unix timestamp in seconds for when the current minute rolls over.
Exceed the budget and you get 429 with a Retry-After header in seconds. Wait that long and continue. There is no penalty for hitting the limit.
Writing a well-behaved client
Read X-RateLimit-Remaining and slow down as it approaches zero, rather than sprinting into a 429 and reacting to it.
When you do get a 429, honour Retry-After instead of retrying immediately.
Prefer one large page over many small ones. A single call with ?limit=200 costs one request; twenty calls at ?limit=10 cost twenty.
X-RateLimit-Degraded
If you ever see X-RateLimit-Degraded: 1, the limiter itself was briefly unavailable and your request was allowed through uncounted.
This is a deliberate choice rather than a bug. A limiter outage that blocked requests instead would look like a total API outage to everyone at once, which is a far worse failure than a short window of uncounted calls. You do not need to do anything when you see this header.