Local MCP (Claude & ChatGPT)
Tool Reference

Tool Reference

These are all 87 tools TimeBolt gives an AI assistant. You don't need to call them yourself: the assistant chooses them from your request. This page is useful when you want to know whether something is possible, or want to name a tool or option in your request.

Kind: Read tools only look at things and change nothing, so AI apps can let them run without asking. Change tools change TimeBolt's project or settings, or write files.

How the assistant sees your project

Timeline format

A timeline is a list of cuts in order, starting at 0 seconds with no gaps between them. The position of a cut in the list is its index, starting from 0. Each cut has:

FieldMeaning
start, durationWhere the cut starts and how long it lasts, in seconds of the original file.
operationkeep, remove or fastforward.
punchedPunch-in zoom, from 1 (no zoom) up to 5.
punchedPositionWhere the zoom is anchored: center, top, bottom, left, right, top-left, top-right, bottom-left or bottom-right.
markedtrue when the cut is marked (starred).
titleA chapter title. A chapter starts at each cut with a title.

Transcription format

A transcription is a list of words, each with its start and end time in seconds:

[
  { "start": 0.12, "end": 0.48, "word": "Hello" },
  { "start": 0.48, "end": 0.90, "word": "everyone" }
]

The position of a word in the list is its word index. You can give the assistant a transcription in this format, made with any other service, instead of transcribing in TimeBolt.

Multi-Track Editor

Tools for the Multi-Track Editor have multitrack in their name. Tracks are numbered from 0, the top track. Their cuts can only be keep or remove, and their times are relative to the start of each track.

Files

File paths are full paths, such as /Users/me/Movies/video.mp4 or C:\Videos\video.mp4. Exported files are saved next to the source file unless a path is given, and a name that is already taken gets a number instead of being overwritten.

App, windows and settings

ToolKindWhat it doesInputs
get_app_stateReadWhat TimeBolt is doing now: version, licence, which windows are open, and the Claude and ChatGPT connection state.none
get_project_stateReadThe project in the main editor: file, loading state, duration, frame rate, resolution, audio tracks, whether it was edited, transcribed or is rendering, and timeline totals (cut counts, kept, removed and fast-forwarded time, estimated output length, markers, chapters).none
open_windowChangeOpens the Multi-Track Editor, Capture (screen recorder), Stitch, File Doctor, Re-encoder, Variable Frame Rate Detector or Settings.window; optional filePath for File Doctor
close_windowChangeCloses one of those windows. Closing Capture stops a recording in progress.window
get_settingsReadTimeBolt's saved settings (File > Settings): audio, render, dark mode, timeline behaviour, punch zoom levels, Umcheck, screen recorder and export options.none
update_settingsChangeChanges only the settings you name. Dark mode applies after a restart.settings

Opening files and silence detection

ToolKindWhat it doesInputs
open_file_in_timelineChangeOpens a video or audio file in the main editor and runs silence detection, like dropping the file onto TimeBolt.filePath; optional replaceOpenFile, detectSilences
get_file_pathReadThe path of the open file.none
get_durationReadThe length of the open file in seconds.none
get_silence_detection_optionsReadThe silence detection and output options under the waveform: filter level, grace period, ignore detections shorter than, padding, fast-forward speed and limits, transitions, processing speed and output quality.none
set_silence_detection_optionsChangeChanges those options without running detection again.options
get_suggested_filter_levelReadThe filter level (dB) TimeBolt would pick automatically for the open file.none
run_silence_detectionChangeRuns silence detection again (the Update Graph button), optionally with new options. Replaces manual edits; undo brings them back.optional options
control_previewChangePlays, pauses, stops or seeks the preview player.action; optional seconds

Timeline editing

ToolKindWhat it doesInputs
get_timelineReadThe current timeline (see Timeline format).none
apply_timelineChangeReplaces the whole timeline after checking it has no gaps or overlaps.timeline
update_cutsChangeChanges the operation, marker, chapter title or zoom of some cuts: by index, by time range, or all of them.indexes, range or all; operation, marked, title, punched, punchedPosition
split_cut_atChangeSplits the cut at a time into two cuts.seconds
merge_consecutive_cutsChangeMerges neighbouring cuts that have the same settings (the Merge Consecutive Cuts button).none
shift_cutsChangeMoves every cut earlier or later by a number of seconds (the Shift Cuts control).seconds
keep_only_marked_cutsChangeRemoves every cut that is not marked.none
undoChangeUndoes the last timeline change, like Cmd/Ctrl + Z.none
save_timeline_jsonChangeSaves the timeline as a TimeBolt .json file (File > Save Timeline).optional savePath
load_timeline_jsonChangeLoads a saved timeline .json, optionally shifted by some seconds.filePath; optional shiftSeconds

Rendering

ToolKindWhat it doesInputs
quick_renderChangeRenders the current timeline right away (the Quick Render button).optional audioOnly, turboSpeedMultiplier, outputPath, burnCaptions
add_to_render_queueChangeAdds the current timeline to the render queue without starting it. Can queue only marked cuts, or one render per chapter.optional audioOnly, turboSpeedMultiplier, markedOnly, chapterClips, outputPath, burnCaptions
get_render_queueReadThe renders waiting in the queue.none
start_renderChangeRenders everything in the queue, one after another (the Start Render button).none
get_render_statusReadWhether TimeBolt is rendering, the progress of the current render, what is still queued and the finished files.none
cancel_renderChangeStops the current render and keeps the rest of the queue.none
remove_from_render_queueChangeRemoves a waiting render from the queue.id
set_background_musicChangeSets or clears the background music mixed under renders, with a volume from 0 to 1.optional filePath, volume, clear
export_cuts_to_directoryChangeRenders each kept cut as its own numbered file in a folder (the Export Cuts button).directory; optional turboSpeedMultiplier
batch_process_filesChangeOpens each file with the current silence detection settings, queues a render for each and starts the queue (Batch Processing).filePaths

Exports for other editors

ToolKindWhat it doesInputs
generate_xmlChangeXML for Premiere Pro and DaVinci Resolve.optional savePath, frameRate (needed for audio files)
generate_xml_sequenceChangeXML for Premiere Pro with the edit as a nested sequence.optional savePath, frameRate (needed for audio files)
generate_fcpxmlChangeFCPXML for Final Cut Pro, in the version set in your settings.optional savePath
generate_fcpxml_multicamChangeMulticam FCPXML for Final Cut Pro.optional savePath
generate_edlChangeAn EDL.optional savePath, frameRate (needed for audio files)
generate_camtasia_projectChangeWrites your cuts into the Camtasia project the video came from.projectPath (.tscproj, or .cmproj on Mac)
generate_srtChangeSRT captions that follow your edited timeline. Needs a transcription.optional savePath
generate_chapters_textChangeA YouTube-style chapter list with timestamps for the rendered video, optionally for a Turbo Mode speed.optional savePath, turboSpeedMultiplier

If an export can't be made, for example when the timeline has fewer than two cuts or a render is running, the assistant gets TimeBolt's reason.

Transcription and filler words (Umcheck)

ToolKindWhat it doesInputs
transcribe_locallyChangeTranscribes the open file on your computer (Umcheck > Local Transcription). The first run downloads the speech model. Replaces any existing transcription.none
get_transcription_statusReadWhether a transcription is running (with progress), ready or failed.none
cancel_transcriptionChangeCancels the running transcription.none
get_umcheck_transcriptionReadThe word-level transcription (see Transcription format).none
set_transcriptionChangeUses a transcription made elsewhere (Umcheck > Apply Transcription File).words or filePath
save_transcription_jsonChangeSaves the transcription as <video>.umcheck.json.optional savePath
find_in_transcriptionReadFinds every place a word or phrase is said, with times.phrase; optional ignoreCase, ignorePunctuation
replace_in_transcriptionChangeReplaces a word or phrase everywhere, for example to fix a misheard name before making captions (Find & Replace All).find, replace; optional ignoreCase, ignorePunctuation
detect_filler_wordsReadLists the filler words it would cut, and optionally false starts (repeated words) and retakes (repeated sentences). Cuts nothing.optional words, falseStarts, retakes
cut_wordsChangeRemoves the given words from the output, like Umcheck's Turn Off Selected Words.wordIndexes
get_predefined_filler_words_and_phrasesReadYour saved filler words and phrases.none
set_filler_words_listChangeReplaces your saved filler words and phrases.words (comma-separated)

Multi-Track Editor

ToolKindWhat it doesInputs
open_files_in_multitrack_editorChangeAdds files as tracks, one per file, opening the Multi-Track Editor if needed.filePaths
get_multitrack_stateReadThe tracks (file, type, length, start position, muted, cuts, transcribed), whether detection has run, and whether it is rendering or busy.none
get_multitrack_timelineReadEvery track with its cuts, optionally with each track's words.optional includeTranscription
apply_multitrack_timelineChangeReplaces the cuts, start positions and mute states of the tracks. Used for camera switching.timeline
synchronize_multitrack_tracksChangeLines up the tracks by their audio, using track 0 as the reference (Synchronize Tracks).none
set_multitrack_track_start_positionChangeMoves one track to start at a given time (Set Track Position).trackId, seconds
swap_multitrack_tracksChangeSwaps two tracks. The video shows the top-most track that is kept at each moment.trackIdA, trackIdB
mute_or_unmute_track_in_multitrack_editorChangeMutes or unmutes a track's audio.trackId, status
remove_multitrack_trackChangeRemoves a track. Cannot be undone.trackId
get_multitrack_silence_detection_optionsReadThe Multi-Track Editor's own silence detection options.none
set_multitrack_silence_detection_optionsChangeChanges those options.options
run_multitrack_silence_detectionChangeRuns silence detection on each track independently, or on one track and copies its cuts to the others (Run Silence Detection). Replaces manual edits.type; optional trackId, keepIfAnyTrackSpeaks
split_multitrack_atChangeSplits every track at a time.seconds
undo_multitrackChangeUndoes the last cut change in the Multi-Track Editor.none
transcribe_multitrack_locallyChangeTranscribes every track that isn't transcribed yet, on your computer.none
get_multitrack_transcription_statusReadTranscription progress, and which tracks are transcribed.none
get_multitrack_transcriptionReadOne track's words.trackId
set_multitrack_transcriptionChangeGives one track a transcription made elsewhere.trackId; words or filePath
start_multitrack_renderChangeRenders the edit to an .mp4 (or .wav when there is no video).renderToPath
get_multitrack_render_statusReadRender progress.none
cancel_multitrack_renderChangeStops the render and deletes the partial file.none
generate_multitrack_fcpxmlChangeFCPXML for Final Cut Pro.optional savePath
generate_multitrack_xmlChangeXML for Premiere Pro and DaVinci Resolve.optional savePath

Capture (screen recorder)

ToolKindWhat it doesInputs
get_screen_recorder_stateReadWhether Capture is open, recording or saving, its settings, and the screens, microphones and cameras available.none
set_screen_recorder_settingsChangeSets the recording type (screen and camera, screen only, camera only), quality, crop (16:9, square or 9:16), screen, microphone, camera and system audio.settings
start_screen_recordingChangeStarts recording after a 3-second countdown. On Mac, screen recording permission must already be granted.none
stop_screen_recordingChangeStops recording and saves an .mp4, which then opens in the main editor.savePath

Media files and conversions

ToolKindWhat it doesInputs
probe_fileReadChecks a file before opening it: whether TimeBolt can edit it (and why not), length, resolution, frame rate, codecs and audio streams.filePath
detect_variable_frame_rateReadChecks a video for variable frame rate, like File Doctor, and suggests a constant frame rate.optional filePath
convert_to_constant_frame_rateChangeConverts a video to a constant frame rate H.264 MP4 (<name>_CFR.mp4), optionally opening the result in TimeBolt.optional filePath, outputPath, fps, openInTimeline
re_encode_fileChangeRe-encodes a file like the Re-encoder: video to H.264 MP4, audio to WAV (<name>_re-encoded).filePath; optional outputPath, fps
get_conversion_statusReadProgress of the latest conversion or re-encode.optional jobId
cancel_conversionChangeCancels the running conversion and deletes the partial file.none