22 commands in the connected MCP tool listing, reviewed September 28, 2026. These are the server commands available to connected AI assistants; tool prefixes vary by client. This reference describes the exposed schemas, not a test of every operation.
ChatGPT and Claude can read and edit timelines, control multitrack projects, mute tracks, render, and export. Connect your AI →
Read the current project
| Command | Input | Purpose |
|---|
get_file_path | {} | Read the open single-file media path. |
get_duration | {} | Read timeline duration; confirm the time basis before using it as edited-output length. |
get_timeline | {} | Read the existing single-file cuts before modifying them. |
get_umcheck_transcription | {} | Read the prepared word-level transcript. An empty result means a transcript must be prepared or supplied. |
get_predefined_filler_words_and_phrases | {} | Read the predefined comma-separated filler list. |
get_render_queue | {} | Read remaining queued render tasks. |
Edit a single-file timeline
| Command | Input | Purpose |
|---|
open_file_in_timeline | filePath: string | Open a media file by its valid path. |
apply_timeline | timeline: string | Apply a serialized JSON array of cuts, including keep/remove/fastforward operations, punch positions, markers, and chapter titles. |
Control a multitrack project
| Command | Input | Purpose |
|---|
open_files_in_multitrack_editor | filePaths: string | Open camera and audio files. Pass a JSON array of valid file paths serialized as a string. |
get_multitrack_timeline | {} | Read the existing tracks and cuts. Inspect this state before editing or camera switching. |
apply_multitrack_timeline | timeline: string | Apply track state and cuts, including source paths, offsets, mute state, and optional word-level transcription. |
mute_or_unmute_track_in_multitrack_editor | trackId: number; status: boolean | Mute or unmute a track. IDs start at 0; true means muted, false means unmuted. |
Render and export
| Command | Input | Purpose |
|---|
add_to_render_queue | {} | Queue the current single-file timeline. |
start_render | {} | Start all queued render tasks. |
start_multitrack_render | renderToPath: string | Render the Multi-Track Editor timeline to the required output path. |
generate_fcpxml | {} | Generate Final Cut Pro XML, save it, and return the path. |
generate_fcpxml_multicam | {} | Generate multicam FCPXML, save it, and return the path. This is distinct from the Multi-Track Editor export below. |
generate_multitrack_fcpxml | savePath: string | Export the Multi-Track Editor project as FCPXML to the required path. |
generate_xml | {} | Generate XML for Premiere Pro or DaVinci Resolve and return its path. |
generate_xml_sequence | {} | Generate nested-sequence XML for Premiere Pro and return its path. |
generate_multitrack_xml | savePath: string | Export XML from the Multi-Track Editor for Premiere Pro or DaVinci Resolve. |
generate_srt | {} | Generate SRT captions from the timeline and transcription, save the file, and return its path. |
Timeline contracts
Both apply commands accept a timeline string containing a JSON array. Read and modify the existing timeline, preserving previous edits. Cover gaps explicitly with remove cuts.
Single-file cuts
Cut members: start, duration, type, punched, punchedPosition, operation, marked, and title. The description defines operation as keep, remove, or fastforward. It names type without specifying a separate enum. Punch scale is at least 1 and less than 5. Positions are center, top-left, top-right, bottom-left, bottom-right, top, bottom, left, or right. marked is boolean; title marks the start and name of a chapter or topic.
Multitrack state
Each track contains id, duration, isMuted, isVideo, sourceUrl, startPosition, timelineCuts, and optional transcription. Duration is source-file length in seconds. startPosition places the track relative to the other tracks. Transcription entries contain start, end, and word.
Each multitrack cut contains start, duration, type, punched, punchedPosition, and operation. The exposed contract allows operation: keep/remove, punched: 1, and punchedPosition: center. Do not assume the single-file punch, fast-forward, marker, or chapter fields work in multitrack.
Camera switching and master audio
For each time range, keep only the selected video angle and mark the other camera angles as removed. When video keep cuts overlap, the preview shows the topmost kept track; track 0 is at the top. Muting controls audio, not camera visibility. Preserve the intended master-audio track and account for track offsets when aligning cuts.
What remains application-side
This tool listing has no dedicated command to start UMCheck, run waveform synchronization, inspect video frames, configure render format, or export multitrack SRT/chapter files. Multitrack transcription and offsets can be represented through apply_multitrack_timeline; that is different from dedicated transcription or synchronization commands. Use UMCheck and synchronization in TimeBolt as needed, then let the assistant work with the prepared project. A file path alone does not establish visual access to camera footage.
Known behaviors and multitrack rules →
