These are the tools available once you've connected Supercut to your AI tool over MCP. For more info, refer to this documentation or see Model Context Protocol (MCP) or Supercut Claude Connector.

Most tools take a public_id: the share ID of a recording or playlist, the last segment of its share link (supercut.ai/share/{workspaceName}/{shareId}).

Lists recordings in the workspace. Choose owned, watched, or shared with the list parameter, defaults to owned. Supports a title search on the owned list, plus cursor and limit for pagination.

Owned recordings include an analytics object with all-time visits and plays counts, so an agent can spot your best-performing recordings without a separate call per recording.

Searches recordings in the workspace by keyword or meaning — matches titles and AI summaries, and understands intent as well as exact wording. Takes a query, plus optional cursor and limit for pagination. Returns recordings you created, were granted access to, or have watched, best match first.

Returns metadata for a single recording: title, owner, duration, processing status, and created date.

Includes the same analytics object when you have access to that recording's analytics.

Returns viewership analytics for a single recording: visits and plays (each with total and unique counts), plus engagement rate (the average percent of the video watched). Pass an optional start and end date (ISO 8601) to scope the numbers to a date range — it covers all time by default.

Fetches the full transcript for a recording.

This is a great way of download the transcript of your videos.

Returns a signed URL for a still frame at a specific timestamp. Pass a time in seconds (decimals allowed), an optional quality tier (uhd, qhd, fhd, hd, sd), and an optional source (screen, camera, or screen_and_camera).

MCP defaults source to screen, so your agent sees what's on screen instead of a frame with the camera bubble in the way. Recordings without separate screen and camera tracks (uploads and imports) fall back to the composed frame automatically. Supercut falls back to the best quality available for the requested tier.

Lists comments and replies left on a recording.

Lists emoji reactions placed directly on a recording's timeline.

Lists viewer form-response submissions for a recording. Supports cursor and limit for pagination.

Lists playlists in the workspace. Supports cursor and limit for pagination.

Returns metadata for a single playlist.

Lists the recordings inside a playlist. Supports cursor and limit for pagination.

Adds a recording to an existing playlist. Requires edit access to the playlist and view access to the recording. Fails if the recording is already in the playlist. This is currently the only MCP tool that writes data. Every other tool is read-only.

Your AI tool can read and organize your Supercut data directly: recordings, transcripts, frames, comments, reactions, form responses, playlists, search, and analytics, all through these tools. New tools get added as MCP grows, and this article will track them as they ship.