exportResponses
ChatGPTExport the existing dashboard response and viewer datasets for a video, image, or channel as XLSX, CSV, or JSON. Applies the same videoDataFilters as dashboard exports; hidden responses are excluded by default where supported by the dataset. Returns a private download link valid for one hour, sheet names and row counts, not inline raw rows. CSV combines sheets with _dataset and _sheet columns. Requires workspace membership. Channel exports support at most 100 active media items per request. Pass teamId when a workspace is established in this conversation. Exact item IDs resolve their workspace automatically. If multiple workspaces are available and no target is established, ask the user using the returned team choices. Team selection applies only to this call.
getAggregations
ChatGPTGet aggregate feedback metrics for a video or feature, including emotions, keywords, sentiment, rating, slider, emoji, and touch breakdowns. viewerPopulation defaults to saved filters; all requires membership. calendarDateRange filters record creation dates; selectedScene is playback seconds. appliedScope reports the effective selection. Pass teamId when a workspace is established in this conversation. Exact item IDs resolve their workspace automatically. If multiple workspaces are available and no target is established, ask the user using the returned team choices. Team selection applies only to this call.
getChannelInsightsTable
ChatGPTGet the aggregate insights-table metrics for a channel and an explicit list of up to 100 member video IDs, including sentiment, ratings, emoji, views, live views, and watch time. Pass teamId when a workspace is established in this conversation. Exact item IDs resolve their workspace automatically. If multiple workspaces are available and no target is established, ask the user using the returned team choices. Team selection applies only to this call.
getChannelVideos
ChatGPTList active videos included in a channel’s sections, with canonical absolute url links for the channel and each video. Use this to establish the video IDs for channel-wide analysis. Pass teamId when a workspace is established in this conversation. Exact item IDs resolve their workspace automatically. If multiple workspaces are available and no target is established, ask the user using the returned team choices. Team selection applies only to this call.
getCollectionItems
ChatGPTList items matching a collection’s saved filters, including each item’s canonical absolute url. Collections are dynamic saved searches, unlike static folders. Use offset when totalItems exceeds the returned item count. Pass teamId when a workspace is established in this conversation. Exact item IDs resolve their workspace automatically. If multiple workspaces are available and no target is established, ask the user using the returned team choices. Team selection applies only to this call.
getContentIntelligence
ChatGPTGet AI-detected entities such as characters, themes, events, and brands for one video. Summary detail is the default; use full only when per-entity moments and timestamps are required. Pass teamId when a workspace is established in this conversation. Exact item IDs resolve their workspace automatically. If multiple workspaces are available and no target is established, ask the user using the returned team choices. Team selection applies only to this call.
getItemByIngestionId
ChatGPTResolve an external ingestion ID to the corresponding video item, Lumiere video ID, and canonical absolute url in the authenticated workspace. Pass teamId when a workspace is established in this conversation. Exact item IDs resolve their workspace automatically. If multiple workspaces are available and no target is established, ask the user using the returned team choices. Team selection applies only to this call.
getItemMetadata
ChatGPTGet selected metadata categories and the canonical absolute url for a Lumiere item ID. Defaults to basic information and hierarchy. Request metrics for view counts; viewerPopulation and calendarDateRange apply to that category, with effective selection in metrics.appliedScope. Other categories are not date-filtered. Analytics tools require a video even though both videos and images use v-prefixed IDs. Pass teamId when a workspace is established in this conversation. Exact item IDs resolve their workspace automatically. If multiple workspaces are available and no target is established, ask the user using the returned team choices. Team selection applies only to this call.
getMedia
ChatGPTRetrieve media as a resource link: video-file (an available rendition, not necessarily the original upload), video-stream (optional segment-aligned excerpt), frame at a timestamp, configured thumbnail/poster, image item asset, or GIF up to 10 seconds. Frame/GIF options include width, height and fit; GIF adds fps. Returns available:false with a reason when absent. Does not include storyboards, transcripts, or analysis. Configured thumbnail/poster images use their stored size. Image items optionally accept a preset width, aspectRatio and resize mode. Proxy or existing HTTPS image URLs may not have a known expiry. URLs expire; request again when needed. Pass teamId when a workspace is established in this conversation. Exact item IDs resolve their workspace automatically. If multiple workspaces are available and no target is established, ask the user using the returned team choices. Team selection applies only to this call.
getMoments
ChatGPTGet AI-generated moments for one video, optionally constrained to time ranges in seconds. A moment is semantic AI analysis; it is different from an authored scene/chapter and from a low-level visual cut. Use pagination and a limit of at most 100. Pass teamId when a workspace is established in this conversation. Exact item IDs resolve their workspace automatically. If multiple workspaces are available and no target is established, ask the user using the returned team choices. Team selection applies only to this call.
getPodVideos
ChatGPTList videos in a Video Pod feature attached to a parent video, with canonical absolute url links for the parent and each child video. A Video Pod is an embedded collection of videos, not a channel. If the parent has multiple pods, use featureId to select one. Pass teamId when a workspace is established in this conversation. Exact item IDs resolve their workspace automatically. If multiple workspaces are available and no target is established, ask the user using the returned team choices. Team selection applies only to this call.
getQuestions
ChatGPTGet responses and aggregations for one interactive question feature on a video. Summary mode is the default. Use full only for requested verbatim or per-response analysis; the response must remain below the MCP output ceiling. Defaults to saved viewer filters and all dates. Use viewerPopulation and calendarDateRange to select the audience and record creation window; appliedScope reports the selection. Pass teamId when a workspace is established in this conversation. Exact item IDs resolve their workspace automatically. If multiple workspaces are available and no target is established, ask the user using the returned team choices. Team selection applies only to this call.
getQuestionsBatch
ChatGPTGet summary responses and aggregations for 1–20 question features on one video in parallel. Each result includes publicationStatus: published, unpublished (including AI suggestions), imported, not_found, or not_question; null on query errors. Check this before interpreting zero responses as a lack of viewer feedback. This MCP tool is summary-only; call the single-question tool in full mode when verbatim responses are required. Defaults to saved viewer filters and all dates. Use viewerPopulation and calendarDateRange to select the audience and record creation window; appliedScope reports the selection. Pass teamId when a workspace is established in this conversation. Exact item IDs resolve their workspace automatically. If multiple workspaces are available and no target is established, ask the user using the returned team choices. Team selection applies only to this call.
getRecruitment
ChatGPTGet recruitment lifecycle and display status, completion target, delivered counts, quotas, targeting configuration, pricing and data freshness. Requires itemId and recruitmentId from listRecruitments. Unknown progress is null; provider refresh failures are identified. Does not launch, pause or purchase recruitment. Pass teamId when a workspace is established in this conversation. Exact item IDs resolve their workspace automatically. If multiple workspaces are available and no target is established, ask the user using the returned team choices. Team selection applies only to this call.
getSceneMetrics
ChatGPTGet exact feedback and retention metrics for authored scene ranges in a video, with precomputed comparisons such as biggest drop-off. A scene is a named time range, not an AI moment or visual cut. Times are seconds; when ranges are omitted, stored video scenes are used. Pass teamId when a workspace is established in this conversation. Exact item IDs resolve their workspace automatically. If multiple workspaces are available and no target is established, ask the user using the returned team choices. Team selection applies only to this call.
getTranscript
ChatGPTRead timestamped transcript pages in JSON or WebVTT. Replaces getCaptions. Supports language, speaker labels/entity IDs, original-language word timings, playback ranges, and pagination. Request fewer utterances for a compact excerpt; follow nextCursor for the full transcript. Missing translations fall back to the original language, reported per utterance. Word timings are omitted for translated text. VTT cue ends are inferred, not word-accurate. All times are seconds. Pass teamId when a workspace is established in this conversation. Exact item IDs resolve their workspace automatically. If multiple workspaces are available and no target is established, ask the user using the returned team choices. Team selection applies only to this call.
getTrendData
ChatGPTRetrieve trend line buckets for a video, matching the Studio insights charts. Supports comment volume, sentiment, rating, slider, emoji, persistent emoji, and retention metrics. Defaults to saved viewer filters and all dates. Use viewerPopulation and calendarDateRange to select the audience and record creation window; appliedScope reports the selection. Buckets and interval remain playback seconds, using the UI granularity by default. Returned viewer filter metadata summarizes the condition and configured ID count without returning respondent IDs. These trends are buckets along the video timeline in seconds, not calendar activity totals. Includes Lumiere trend charts by default. Use presentation:none for intermediate analysis or text-only requests. Do not call presentResults again for an auto-rendered trend. Pass teamId when a workspace is established in this conversation. Exact item IDs resolve their workspace automatically. If multiple workspaces are available and no target is established, ask the user using the returned team choices. Team selection applies only to this call.
listItems
ChatGPTBrowse workspace items without a search term. Replaces getFolderContents. Defaults to direct children of ROOT; set recursive for bounded traversal through folders, channels, and pods. Results are flat with parent IDs. Continue with cursor even when a page is empty. Type filters affect results, not traversal. depthLimited indicates descendants were omitted by maxDepth. Includes interactive item cards by default (presentation: auto). Use presentation: none for data-only lookups or explicit text-only requests; do not call presentResults again for the same listing. Item results include a canonical absolute url that opens the item in Lumiere. Pass teamId when a workspace is established in this conversation. Exact item IDs resolve their workspace automatically. If multiple workspaces are available and no target is established, ask the user using the returned team choices. Team selection applies only to this call.
listRecentItems
ChatGPTList the most recently created or edited active folders, videos, or channels in one workspace. Use itemType folder and limit 1 to find the newest folder, then listItems. orderBy is required: clarify whether recent means created or edited when unclear. This does not track what the user last opened or filter to items created by that user. The search index may briefly lag new changes. Includes interactive item cards by default (presentation: auto). Use presentation: none for data-only lookups or explicit text-only requests; do not call presentResults again for the same listing. Item results include a canonical absolute url that opens the item in Lumiere. Pass teamId when a workspace is established in this conversation. Exact item IDs resolve their workspace automatically. If multiple workspaces are available and no target is established, ask the user using the returned team choices. Team selection applies only to this call.
listRecruitments
ChatGPTList recruitments across the authenticated team, optionally restricted to one item. Defaults to Live, Paused and Auto-Paused; supply statuses to include drafts or completed/cancelled recruitment. Bounded pagination can return an empty page with a nextCursor: continue until nextCursor is null. Status filters use refreshed data when available; inspect freshness for fallback data. Pass teamId when a workspace is established in this conversation. Exact item IDs resolve their workspace automatically. If multiple workspaces are available and no target is established, ask the user using the returned team choices. Team selection applies only to this call.
listTeamActivity
ChatGPTPage through team content, responses, engaged-videos, or viewer-sessions. last accepts 30m, 12h, or 7d; alternatively use from/to ISO date-times. Content defaults to videos ordered by editedAt; for newly created content including folders, set contentType: all and contentOrder: createdAt. Use from/to with timezone offsets for a calendar week. Sessions sort by session start or last engagement. Engaged videos sort by itemId and report the latest matching viewerOrder timestamp plus exact visit-document sessionCount (not unique viewers). Follow nextCursor with the same filters and returned pagination.limit until completeness.complete is true; all pages must finish within ten minutes of the first request, including after empty pages. Replays do not extend that deadline; at most 20 snapshots may be reserved per team, with capacity held for up to twenty minutes while ES releases resources. Counts cover the search snapshot; totalEntries becomes available after traversal and excludes moved/deleted items. Responses return at most 20 records per page, sentiment and slider values, and 280-character previews with explicit textTruncated. For counts, set responseText=none. To retrieve a full comment, set responseId to its id, then repeat with nextTextOffset as textOffset until null (up to 4000 UTF-16 units per call, without splitting surrogate pairs). Details ignore time bounds, read current data without a feed cursor, and return responseFound: false if unavailable. Public callers cannot use this tool. Pass teamId when a workspace is established in this conversation. Exact item IDs resolve their workspace automatically. If multiple workspaces are available and no target is established, ask the user using the returned team choices. Team selection applies only to this call.
listTeams
ChatGPTList workspaces authorized for this connection where you are still an active member. Returns IDs and names. Use a workspace ID on subsequent calls when the user establishes a workspace in this conversation. No workspace selection is stored on the server. For explicit cross-workspace discovery, search each listed workspace and label results. Reconnect to authorize additional workspaces.
presentResults
ChatGPTUse this when displaying a standalone data trend or comparison with Lumiere styling, or arranging a custom selection of Lumiere item cards. Prefer trend-chart for a supported standalone trend and comparison-chart for comparing one metric across videos or categories; use another presentation approach for complex reports or unsupported visualizations. searchItems, listItems, listRecentItems and getTrendData already render by default: do not repeat their auto presentation. Supports items-list (ordered itemIds) and trend-chart (title, xAxis:date|seconds|number, optional xDomain/axis labels/unit/sourceItemIds, and 1–4 series with unique id, label, and points:{x,y}). date x values are ISO dates or UTC timestamps; other axes use numbers. Points must increase strictly in x. Optional xDomain:[min,max] uses the same type as x, must contain every point, and keeps the full timeline visible. When charting video buckets from getTrendData, use midpoint x=min(bucket key + interval/2, duration.max) and xDomain:[duration.min,duration.max], matching Studio. comparison-chart uses title, optional valueLabel/unit/sourceItemIds, 1–20 categories:{id,label}, and 1–4 series:{id,label,values:[{categoryId,value}]}. Each series must supply exactly one value per category; use null for missing values. IDs must be unique. All series must use the same metric and unit; use separate components for unlike metrics. Retrieve the evidence first; when using getTrendData as input for a comparison, request presentation:none to avoid intermediate charts. Category order is preserved. Bars share a zero baseline and support negative values. Use null for missing data, never fabricate values or IDs. Supplied series are labeled as provided data, not verified Lumiere analytics. All referenced items must be active in one authorized workspace. At most 4 components, 50 items and 2000 trend points total; at most 500 points per trend series and 80 values per comparison chart. Do not supply HTML, scripts, arbitrary styles, card metadata or URLs. Pass teamId when a workspace is established in this conversation. Exact item IDs resolve their workspace automatically. If multiple workspaces are available and no target is established, ask the user using the returned team choices. Team selection applies only to this call.
searchItems
ChatGPTSearch the authenticated workspace for videos, channels, or folders by name. Each item includes a canonical absolute url that opens it in Lumiere. Includes interactive cards by default (presentation: auto); use presentation: none for intermediate ID lookups or text-only requests. Do not call presentResults again for the same listing. A v-prefixed item can be a video or image; inspect the returned type instead of inferring media type from the ID. Pass teamId when a workspace is established in this conversation. Exact item IDs resolve their workspace automatically. If multiple workspaces are available and no target is established, ask the user using the returned team choices. Team selection applies only to this call.
searchKnowledgeBase
ChatGPTSearch Lumiere’s published help articles for product features, settings, and workflows. Returns article markdown and absolute links to the relevant knowledge-base pages. An empty articles array means no relevant published guidance was found; say so instead of inventing instructions.
semanticMomentSearch
ChatGPTSemantically search AI-generated moments across the authenticated workspace. Use this for concept or meaning-based discovery across videos; results are moments, not authored scenes or visual cuts. Pass teamId when a workspace is established in this conversation. Exact item IDs resolve their workspace automatically. If multiple workspaces are available and no target is established, ask the user using the returned team choices. Team selection applies only to this call.
getAggregations
ClaudegetCaptions
ClaudegetChannelInsightsTable
ClaudegetChannelVideos
ClaudegetCollectionItems
ClaudegetContentIntelligence
ClaudegetFolderContents
ClaudegetItemByIngestionId
ClaudegetItemMetadata
ClaudegetMoments
ClaudegetPodVideos
ClaudegetQuestions
ClaudegetQuestionsBatch
ClaudegetSceneMetrics
ClaudegetTrendData
ClaudesearchItems
ClaudesearchKnowledgeBase
ClaudesemanticMomentSearch
Claude
