data-explorer

name	data-explorer
description	Guides advanced data querying, filtering, and export workflows. Use when users want to browse content, filter by specific criteria, export data, or perform custom analysis beyond standard insights. Triggers on: "export", "download", "filter", "show me items", "browse", "search for", "find content", "CSV", "gallery".

Data Exploration and Export

Your primary lever for controlling what you get back from every tool is field selection. Master fields and labelFields before anything else — they determine whether you get a useful, focused response or a wall of noise.

Field Selection (Start Here)

All item-returning tools (get_items, search_items, analyze, get_item_detail) accept two parameters that control what comes back:

• fields — Which item-level properties to include. Default base set: id, brand, platform, headline, engagement metrics, isOutlier, likesMultiple, createdAt, adDescription, transcript. Add more as needed (e.g., fields: ["hookMechanism", "creativeConcept", "viewsMultiple"]). Pass fields: [] to suppress all item fields (stats-only mode).
• labelFields — Which label dimensions to include on each item. Pass specific categories (e.g., labelFields: ["HookType", "ContentFormat"]) or labelFields: [] to suppress labels entirely.

Use list_labels to discover available dimensions. Before requesting label fields, call list_labels on the knowledge set. It returns every label category and its values — this tells you what dimensions exist so you don't guess at field names.

Never use compact: true — use fields: [] and labelFields: [] instead for the same effect with precise control. compact is a legacy toggle superseded by field selection.

See the available fields reference for the full 45+ field catalog, sort options, and filter formats.

Browsing Content

get_items provides paginated browsing with filters. search_items handles semantic/natural-language queries. Each item can include:

• Identification — id, brand, feedType, platform, sourceLabel
• Content — headline (first 200 chars of post text), transcript, adDescription, url
• Engagement — likes, views, shares, comments, engagementScore, engagementRate
• Performance — timesAccountAvg, likesMultiple, viewsMultiple, commentsMultiple, sharesMultiple, longevityMultiple, isOutlier, outlierType
• Source Baselines — sourceItemCount, sourceAvgLikes, sourceMedianLikes, sourceStddevLikes, sourceP90Likes, sourceAvgViews, sourceMedianViews, sourceAvgComments, sourceMedianComments, sourceAvgShares, sourceMedianShares, sourceAvgDaysAlive
• Media — thumbnail URL, mediaType (image/video)
• Timing — createdAt (ISO date string)

Filters

• Platform — tiktok, instagram, youtube, facebook, twitter, reddit, linkedin, threads
• Feed filter — Narrow to specific brand feeds, influencer feeds, search feeds, or discussion feeds
• Date range — ISO date strings for start/end
• Media format — image, video, or both

Parallel Browsing

When comparing content across different dimensions, run multiple get_items calls in parallel with different filters. Examples:

• Brand A's TikTok content + Brand A's Instagram content (platform comparison)
• Brand A's outliers + Brand B's outliers (competitive comparison)
• Last 30 days + previous 30 days (trend detection)

Each call is independent — do not wait for one to finish before starting the next.

Cross-KS Content Search

content_intelligence_search searches the entire Adology database using AI-powered aspect decomposition — not scoped to any Knowledge Set. Use it when:

• The user wants inspiration beyond their tracked brands
• You need examples of a specific creative approach across the market
• The user's KS doesn't have enough data on a topic

Results include full performance enrichment (multiples, outlier flags, source baselines).

Saving and Sharing

save_to_collection organizes items into named collections. The response includes a url field — a shareable gallery deep link to the collection in the Adology app. Use this whenever you find noteworthy content during browsing or analysis:

• Competitive examples worth revisiting
• Inspiration references for a creative brief
• Top performers to share with a team

Always share the returned URL so the user can view the collection directly.

Export

Bulk data export is available through the Adology web UI. Use save_to_collection to curate items into shareable collections when the user needs to share specific findings.

Analysis Patterns

• Stats-only — get_table_data for aggregate numbers without individual items
• Content-first — get_items with filters or search_items for semantic queries
• Full pipeline — analyze returns sampled items + tableData + stats; drill into standouts with get_item_detail
• Dimension discovery — list_labels to see what label categories exist before filtering or requesting label fields

Error Recovery

• 0 results from get_items — Broaden filters: remove platform constraint, widen date range, drop feed filter. If the KS is new, check get_knowledge_set for item counts — data may not have been scraped yet.
• list_labels returns empty — The KS has items but they haven't been analyzed yet. Fall back to field-based browsing (engagement metrics, platform) without label dimensions.
• Unknown label category — Always call list_labels first rather than guessing category names. The available categories vary by KS depending on which analysis pipeline ran.