Cumulative keyword performance across the raw -> prefilter -> validated funnel.
raw_count descending. Each row aggregates every scan ever run for that keyword — use getKeywordRuns to see the per-run time series.
signal_type_id — narrow to one signal type.source_id — narrow to one adapter (e.g. linkedin_jobs, reddit).offset defaults to 0, limit defaults to 100 and is capped at 500. The response sets nextOffset only when more rows exist.
raw_count then prefilter_count then validated_count. Drops between stages identify where keywords are losing value — large raw-to-prefilter drops suggest noisy keywords; large prefilter-to-validated drops suggest the classifier is rejecting on-topic results (bad disqualifiers or prompt).Safe to poll. No side effects. Use during signal tuning to watch the effect of prompt or keyword changes.Chained call. Click into getKeywordRuns with a stat id to see the per-run snapshot history (last 100 runs, newest first).Bearer authentication header of the form Bearer <token>, where <token> is your auth token.
Filter stats to a single signal type.
^([0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[1-8][0-9a-fA-F]{3}-[89abAB][0-9a-fA-F]{3}-[0-9a-fA-F]{12}|00000000-0000-0000-0000-000000000000|ffffffff-ffff-ffff-ffff-ffffffffffff)$Filter stats to a single adapter (e.g. linkedin_jobs).
Page offset. Defaults to 0.
0 <= x <= 9007199254740991Page size. Defaults to 100, capped at 500.
1 <= x <= 500