Scan a selection for posts that stand out
Requires posts:read. One row per post across a selection (up to
1,000 posts), compared with the selection’s own baseline — the
comment-weighted rollup of the same posts, the same math as the
dashboard’s combined-by-account rows. Every number is a stored post
score; nothing is rescored and no comments are loaded.
Posts with at least 10 analyzed comments get flags (posts under 10
are lowSample: true with no flags). meta.flagRules spells out
each rule:
negative_sentiment: sentiment under 4.5below_baseline/above_baseline: sentiment 1.5+ below / above the baselinecommenters_arguing: 25%+ of comments in the debating or fighting consensus bands — commenters arguing with each other, whatever they think of the postoff_topic: relevance under 4.5content_topic_gap: content and topic scores 2+ points aparthigh_engagement_negative: engagement rate in the selection’s top quarter while sentiment is under 4.5
Flags say where to look; POST /analyses on those posts says why.
flags in the body keeps only posts with any of the listed flags.
engagement.engagementRate is the stored rate, or (likes +
comments + shares + saves) / views when none is stored.
JSON pages with limit / cursor (each page re-resolves the
selection). format=csv returns every row in one call.
Rate class posts_scan (20 / minute, 500 / day per key).
Authentication
Bearer authentication of the form Bearer <token>, where token is your auth token.
Request
Which content the analysis covers. Provide exactly one of postIds
or posts; collectionId may stand alone or be combined with posts
filters to narrow within the collection. Resolves to at most 1,000
posts (newest first).
A field name, or { by, order }. Default postedAt descending; other fields default ascending (worst first for scores).
Response
Scan rows (JSON, paged) or every row as CSV
