recent-tweet-search now describes the repeat job in operational terms. Save the query, language, and max_results values with the post IDs from a response. Run the same request again, then compare IDs and timestamps to find what appeared since the prior check.

That works for a brand mention watch, a competitor query, or a campaign term. The route returns public post text, timestamps, metrics, and expanded author profiles. It stays read-only.

Older Twitter query syntax still works. The route normalizes since:, until:, and -filter:retweets before it asks X for recent results.

Sometimes native X search is unavailable. In that case the response keeps the same tweets array and marks public-index results with degraded: true. A monitor can keep that result separate from a native X result instead of treating an empty response as success.

The request fields, response shape, and price are unchanged.