I've seen another link checking tool mention response caching:

URL request results will be cached when true. This will ensure that each unique URL will only be checked once.

Which is also paired with a cache age option:

Default Value: 3_600_000 (1 hour)
The number of milliseconds in which a cached response should be considered valid. This is only relevant if the cacheResponses option is enabled.

I don't see it documented here if responses are cached by default implicitly. If not, is this something that could help with

I did see this discussion about an extractor pool with the benefit of de-duping URL checks. That's a similar benefit I guess, although if there were a related cache data file that could be persisted (with an expiry timestamp), one could persist this with CI cache tooling across runs.

That would be useful for PRs triggering tests each time commits are pushed to update the PR, reducing concerns about Rate Limits being hit? (since the likelihood of a URL, at least ignoring internal ones if local filesystem URI were later supported; would probably still be successful within a given duration between runs)