get_metric_summary
ChatGPTGet aggregated metrics with optional breakdowns for the filtered dataset. - breakdown_by: Dimensions to break down by. Each is fetched in parallel. Options: source_platform | sentiment | region | gender | age_group | ad_prediction | query_id Note: source_platform = channel/platform breakdown (Instagram, TikTok, YouTube, etc.) - filters: scope the dataset (query_ids, platforms, sentiments, date range, etc.)
get_period_change
ChatGPTGet time-series metrics to analyze volume and sentiment changes over time. When to use: "Why did mentions spike?", "How did the campaign change engagement?", "MoM/WoW trend", "Baseline vs current comparison", "Detect anomaly dates." - interval: DAY | WEEK | MONTH - metric: POST | IMPRESSION | ENGAGEMENT | AVG_ENGAGEMENT | VIEW | LIKE | COMMENT | SHARE | POSITIVE_SENTIMENT | NEGATIVE_SENTIMENT | NEUTRAL_SENTIMENT | NET_SENTIMENT - limit: maximum periods to return (defaults to 14) - Returns each period's post count, engagement, sentiment split, Δ vs previous period, and overall delta. Spike dates (>1.5x average) are flagged automatically.
get_post_details
ChatGPTGet full content, transcript, and metadata of a specific post. The returned post exposes source_platform and the exact text tokens collect_count=<base-10 integer>, subscribe_count=<base-10 integer>. Both tokens are always present, including when either value is 0. Post ID is echoed only to cite the post explicitly selected by the user.
get_post_features
ChatGPTGet AI-extracted features (hooks, content strategy, demographics) for a post. Post ID is echoed only to cite the post explicitly selected by the user.
get_posts_by_ids
ChatGPTBatch fetch post metadata for a list of PostIDs in one call. When to use: After search_voc returns PostIDs and you need author, engagement metrics, platform, date, and content summary for all of them — without calling get_post_details N times. Each returned post exposes source_platform and the exact text tokens collect_count=<base-10 integer>, subscribe_count=<base-10 integer>. Both tokens are always present, including when either value is 0. Each returned Post ID supports a requested drilldown or citation through get_post_details or get_post_features.
get_ranked_posts
ChatGPTGet posts ranked by a performance metric. When to use: "Top performing content", "Worst performing posts", "Rising viral posts", "Best hook examples", "Highest view count in last 30 days." Supported sort_by options: - "DATE": Sort by publish date - "VIEW_COUNT": Sort by number of views - "LIKE_COUNT": Sort by number of likes - "COMMENT_COUNT": Sort by number of comments - "SHARE_COUNT": Sort by number of shares - "ENGAGEMENT_RATE": Sort by engagement rate (engagement / views) - "ENGAGEMENT_TOTAL": Sort by total engagement actions (likes + comments + etc.) - "ENGAGEMENT_IMPRESSION": Sort by engagement impressions Each returned post exposes source_platform and the exact text tokens collect_count=<base-10 integer>, subscribe_count=<base-10 integer>. Both tokens are always present, including when either value is 0. A returned PostID supports a requested drilldown or citation through get_posts_by_ids, get_post_details, or get_post_features. Next Cursor is returned only when another page of the same ranking is available.
get_top_influencers
ChatGPTGet top influencers (authors) in this workspace ranked by a performance metric. It returns display names, public profile URLs, and platform labels, not an internal author identifier. When to use: "Who are the most influential creators/influencers?", "가장 영향력 있는 크리에이터/인플루언서가 누구야?", "Top authors by post count or engagement." Supported sort_by options: - "FOLLOWER_COUNT": Sort by follower count (default) - "POST_COUNT": Sort by number of posts - "ENGAGEMENT_TOTAL": Sort by total engagement actions
get_top_terms
ChatGPTGet top keywords, hashtags, and mentions ranked by frequency. When to use: "What topics are most discussed?", "Top hashtags", "Frequent brand mentions", "USP gaps", "What words appear most in negative posts?" - target_types: List of term types to fetch in parallel. Options: keyword | hashtag | mention (defaults to all three if omitted) - limit: Top N per type.
list_data_queries
ChatGPTList all search queries that collected posts in this workspace. MANDATORY starting point for any analysis to identify relevant query_ids. queries[*].query_id is returned with its query title only to scope a later analysis call through filters.query_ids. Each query has a type that determines the appropriate analysis angle: - BRAND_HEALTH: sentiment trend, SOV, brand mention volume - COMPETITIVE_ANALYSIS: brand comparison, positioning diff, share of voice - CAMPAIGN_EVENT: before/after change, spike detection, campaign reach - INDUSTRY_INSIGHTS: topic clustering, trend discovery, keyword landscape - CUSTOM_TOPIC: user-defined exploration, open-ended VOC - ACCOUNT_TRACKING: per-account performance, creator analysis
list_workspaces
ChatGPTList authorized workspaces and their data queries. workspaces[*].workspace_id is returned only when multiple authorized workspaces require explicit selection for a later tool call. queries[*].query_id is returned with its query title only to scope a later analysis call through filters.query_ids.
search_posts
ChatGPTKeyword-based post collection with quantitative filters. When to use: The keyword is known and you need precise, filtered post collection — "Find posts mentioning 'banapresso'", "Posts about collagen", date range, platform/sentiment filtering. - text: Full-text search against tokenized lexemes (title, content, hashtags, mentions, transcript, comments). Uses websearch_to_tsquery — supports quoted phrases and operators (e.g. '"skin barrier" -dryness'). Only returns posts where the term is indexed in the post's content. - filters.view/like/comment_count_gte/lte: Quantitative engagement filters. - filters.engagement_rate_gte/lte: Filter by engagement rate (e.g. 0.05 = 5%). Use search_voc when the concept is known but exact wording is uncertain. Each returned post exposes source_platform and the exact text tokens collect_count=<base-10 integer>, subscribe_count=<base-10 integer>. Both tokens are always present, including when either value is 0. A returned PostID supports a requested drilldown or citation through get_posts_by_ids, get_post_details, or get_post_features. Next Cursor is returned only when another page of the same search is available.
search_posts_by_summary
ChatGPTSemantic search over posts using post-level summary embeddings. When to use: Find posts whose overall summary is semantically similar to the query — "Posts about skincare routines", "Content expressing purchase intent", "Posts reviewing coffee taste." Unlike search_voc (block-level), this searches the full-post summary — better for post-level discovery. Each returned post exposes source_platform and the exact text tokens collect_count=<base-10 integer>, subscribe_count=<base-10 integer>. Both tokens are always present, including when either value is 0. A returned PostID supports a requested drilldown or citation through get_posts_by_ids, get_post_details, or get_post_features.
search_voc
ChatGPTHybrid search (semantic + keyword boost) over post content blocks to find relevant VOC. When to use: Open-ended, concept-based questions — "What are people saying about X?", "What complaints appear?", "Any purchase motivations?" Keyword match gives a score boost but is not required — semantically similar content without the exact keyword still appears in results. DO NOT use when exact keyword filtering is needed — use search_posts(text=...) instead. A returned PostID supports a requested drilldown or citation through get_posts_by_ids, get_post_details, or get_post_features.
get_metric_summary
Claudeget_period_change
Claudeget_post_details
Claudeget_post_features
Claudeget_posts_by_ids
Claudeget_ranked_posts
Claudeget_top_influencers
Claudeget_top_terms
Claudelist_data_queries
Claudelist_workspaces
Claudesearch_posts
Claudesearch_posts_by_summary
Claudesearch_voc
Claude
