add_adjustment_layer
ChatGPTAdd an adjustment layer (effects apply to everything below) at start_ms for duration_ms. Add effects with add_effect.
add_clip
ChatGPTPlace a source on the timeline at start_ms. Video sources also get a linked audio clip on the matching audio track (with_audio). Fails if the span is already occupied. One undo step.
add_effect
ChatGPTAdd an effect (see list_effect_types) to one or more clips with optional params. Returns effect ids per clip. One undo step.
add_marker
ChatGPTAdd a timeline marker at at_ms with a label and optional color (#RRGGBB). One undo step.
add_mask
ChatGPTAdd a rectangle or ellipse mask to a clip with optional feather, opacity, inverted, expansion. Returns mask_id.
add_nested_sequence
ChatGPTPlace another sequence of the project as a nested clip (sequence_id or sequence_index) at start_ms on a free video track.
add_session_marker
ChatGPTAdd a marker to an analysis session at at_ms (optional end_ms range, label, type, description).
add_shorts_layer
ChatGPTAdd a layer: text (text or binding key), solid, image (path), or broll (empty slot track). Returns clip_id/track_id.
add_solid
ChatGPTAdd a solid color clip (#RRGGBB) at start_ms for duration_ms, default sequence size.
add_text
ChatGPTAdd a text clip (title/caption) on the timeline at start_ms (default: playhead) for duration_ms (default 5000). x/y are the center in 0-1 screen fractions, size is the box height as a fraction of frame height. One undo step.
add_to_pick_board
ChatGPTAdd a time range of a session to its pick board (shorts candidates).
add_track
ChatGPTAdd a video or audio track. Returns track_id.
add_transition
ChatGPTPut a transition on a clip's in or out edge (type from list_effect_types, default crossfade, duration_ms default 500). One undo step.
analyze_scenes
ChatGPTSplit a session into scenes (local, no credits) using transcript + chat + visual cuts. Returns scenes with start/end, title, fun_score, type.
apply_anim_preset
ChatGPTApply an animation preset to a clip with optional params, start_ms/end_ms (timeline), seed; bake=true writes keyframes.
apply_shorts_style
ChatGPTApply a layout style to the active sequence (or sequence_index) of a shorts project. Reversible with undo.
apply_talk_state
ChatGPTMulticam-style switch: from at_ms (default: playhead) to the end, show the given state's image on the speaker's clips (splits the clip under at_ms). state = idle, speaking, join, leave or custom:<key>. Same as clicking a state card in the Talk Scene panel. One undo step.
apply_text_preset
ChatGPTApply a text style / subtitle bar preset (id or name) in bulk. Without clip_ids it applies to every text/caption track in the active sequence; with clip_ids to the tracks of those clips (only_clip=true: just those clips). One undo step.
ask_session
ChatGPTAnswer a natural-language question about one broadcast using its subtitles and chat; returns the answer and time hits. 크레딧이 차감됩니다.
cancel_export
ChatGPTStop the running export. The partial output file is not usable.
cleanup_empty_tracks
ChatGPTRemove empty tracks (keeps the minimum set). Reversible with undo.
clear_keyframes
ChatGPTRemove all keyframes of a clip, or only of one property. Reversible with undo.
clear_masks
ChatGPTRemove all masks from a clip. Reversible with undo.
close_gaps
ChatGPTRemove all empty gaps on one track or on every track, shifting clips left. Reversible with undo.
close_subtitle_designer
ChatGPTClose the subtitle bar designer window without saving. Returns closed=false if no designer window is open.
create_longform_edit
ChatGPTMake a transcript-based long-form edit from pick board entries (main footage + radio-style talk scene) in a new project, opened in the editor. No credits.
create_project
ChatGPTCreate a new empty project and open it in the editor. Defaults: 1920x1080 at 30 fps.
create_sequence
ChatGPTAdd a new sequence to the open project and make it active. Defaults: 1920x1080 at 30 fps.
create_shorts
ChatGPTCut shorts from pick board entries (all picks, or pick_ids): one vertical sequence per pick in a new project, opened in the editor. target_sec -1 = auto length. style_id/style1 applies a layout style. Background per job; poll get_shorts_batch_status. 크레딧이 차감됩니다.
create_talk_scene
ChatGPTBuild a talk-scene sequence (Discord speaker tracks) for the whole session in a new project. Needs recorded speaker tracks.
cut_longform_silence
ChatGPTDetect silences on the long-form audio track (0.7 s, 150 ms margins) and cut them from all tracks. dry_run lists them. Reversible with undo.
cut_silence
ChatGPTDetect silence on an audio track (default: the bottom audio track with clips) and cut every silent range out of all tracks. min_ms = minimum silence length, threshold 0.05-0.95 (lower = stricter). dry_run only lists the ranges. Takes up to a minute. Reversible with undo.
delete_clips
ChatGPTRemove clips (and their linked audio) leaving gaps. Reversible with undo.
delete_pick
ChatGPTRemove a pick board entry. Cannot be undone.
delete_project
ChatGPTPermanently delete a project that is not open (and its default project file). Cannot be undone.
delete_session_marker
ChatGPTDelete an analysis session marker. Cannot be undone.
delete_shorts_layer
ChatGPTDelete a layer by clip_id (whole_layer=true removes its track) or remove a B-roll slot by track_id. Reversible with undo.
delete_transcript_range
ChatGPTRipple-delete a timeline range [start_ms, end_ms) across all tracks (transcript-edit style). Reversible with undo.
delete_transcript_words
ChatGPTDelete words (word_indices from get_transcript with include_words) from a segment; the matching video/audio range is cut out of the timeline. Reversible with undo.
diff_versions
ChatGPTWhat changed between a version and now (or between two versions), grouped by sequence.
duplicate_clip
ChatGPTCopy a clip right after the original on the same track (or onto track_id). One undo step.
edit_transcript_text
ChatGPTReplace the text of one speech segment (segment_index from get_transcript). One undo step.
export_subtitles
ChatGPTWrite the text clips of a track as SRT or YTT to output_path.
find_clips
ChatGPTFind clips by track, name substring, or time range (clips overlapping from_ms..to_ms). Sorted by start_ms.
generate_code_motion
ChatGPTWrite scene code from a natural-language prompt (e.g. '휴대폰 채팅 장면: 말풍선 3개가 차례로 뜨고 알림 배너가 내려온다'), run it, and insert it at start_ms as a code-motion composition. Returns the code and clip ids. 크레딧이 차감됩니다. One undo step.
get_auto_production_status
ChatGPTSteps (chat, voice, outline, highlight, edit) with pending | running | done | failed, project_id, recent log.
get_chat_collect_status
ChatGPTStatus of the chat collection job: running | done | error | not_started, fetched count, chat_count.
get_chat_heatmap
ChatGPTChat message counts per bucket_ms (default 60000) for a session: where viewers reacted most.
get_chat_peaks
ChatGPTTop chat-activity peaks of a session (bucket_ms default 10 s, top_k default 10).
get_clip
ChatGPTOne clip in full detail (including effects) plus its source.
get_code_motion
ChatGPTRead one Code Motion scene: its full code, declared params (name, type number|frame|color|text|textarea|enum|bool, current value, default, min/max/step, options), duration_frames, duration_override_frames (0 = length from the code), fps, width/height, speed, background, animate, text nodes, and the timeline clip_ids that use it. Pick the scene by clip_id or sequence_id (from list_code_motions); with neither, the selected clip or the only scene is used.
get_code_motion_manual
ChatGPTThe manual for writing Strelizia code-motion scene code (phone UI, chat bubbles, typing, notifications, banners as animated layers): primitives, props, param(), timing helpers, examples, installed fonts and the active sequence's fps/size. Read it before run_code_motion.
get_export_status
ChatGPTWhether an export is running, percent done, frames, and any error.
get_frame
ChatGPTRender-accurate PNG of the program monitor at at_ms (seeks there; default: current playhead), downscaled to max_width (default 1280). Use it to check composition, text placement and timing.
get_layer_board
ChatGPTLayer board (shorts panel) state of the active sequence: layers with ids, bindings, hidden flags, title/subtitle/layout state, binding keys.
get_pipeline_status
ChatGPTStatus of the running/last subtitle pipeline: per-stage status, done/total, recent log lines.
get_project_summary
ChatGPTSummary of the open project: active sequence settings, sequence list, track/clip/source counts, playhead.
get_review_room
ChatGPTOne review room with its versions.
get_session_markers
ChatGPTMarkers of an analysis session (label, type, at_ms, in/out range).
get_session_player
ChatGPTWhich analysis session is open in the app and its player position/playing state.
get_session_report
ChatGPTThe cached report (highlights, shorts candidates) of a session, if one was generated in the app.
get_session_thumbnail
ChatGPTSmall JPEG frame of a session's video at at_ms (default: 10% into the video), max_width default 320. For previews; use get_frame for the editor's program monitor.
get_shorts_batch_status
ChatGPTJobs of the current shorts/long-form batch: waiting | running | done | error, sequence_index, project_id, all_done.
get_subtitle_bar
ChatGPTRead the caption bar / text style of a caption or text clip as an editable JSON document. Returns payload (edit it and send it to set_subtitle_bar), style (the clip's current flat text style), kind (subtitle_bar | text_style) and bar_scope (track | clip | none). A subtitle_bar payload is {version:4, canvas:{w,h}, layers:[...bottom to top], inAnim, outAnim, sfxIn, sfxOut}; each layer has id, kind (text|shape|image), name, content, binding ('subtitle' = caption body, 'speaker' = speaker name, '' = fixed text), hidden, tx (normalized position/scale/rotation/opacity), text {family,size,weight,color,stroke,box (background color/radius/padding),align,...}, shape {type,fillColor,fillOpacity,cornerRadius,stroke*}, effects (layer styles such as dropShadow), parentId, anchor, follow (shape wraps a text layer with padL/padR/padT/padB). A text_style payload is the flat style: family,size,weight,color,strokeEnabled/Color/Width,boxEnabled/Color/Radius/PadL..B,align,lineHeight,effects. Sizes are px on canvas.
get_talk_scene
ChatGPTRead the open Talk Scene project: speakers (member id, name, matched person, idle/speaking/join/leave image file names, custom states, frame/ring outline, clip count, track, position/scale, visibility, own audio, captions, sync/pad/min timing), known persons for matching, background and scene config. Call before any set_talk_* tool.
get_timeline
ChatGPTFull state of the active sequence: tracks with their clips (ids, start_ms/end_ms, source_in_ms, speed, volume, opacity, blend, links), sources, text overlays, markers, playhead. Call this before and after editing.
get_transcript
ChatGPTSpeech segments (text, speaker, source start_ms/end_ms, timeline_start_ms) inside a clip that came from an analysis session. include_words adds per-word timing (needed for delete_transcript_words).
get_transcription_status
ChatGPTProgress of start_transcription for a session: running | done | error | not_started, progress %, segment_count.
get_usage
ChatGPTCredits (plan, monthly remaining/used, extra) and cloud storage usage of the signed-in account.
get_video_outline
ChatGPTChapter list (start_ms, title) generated for a project (default: the open project).
import_media
ChatGPTImport a local media file (absolute path on the user's machine) as a source. Returns the source id. Does not place a clip; use add_clip. Importing the same path twice returns the existing source.
import_subtitles
ChatGPTImport an SRT/VTT file as text clips on a track (new video track by default).
import_video_session
ChatGPTCreate a new analysis session from a local video file (probes duration, registers multi-track audio). Returns session_id.
insert_template
ChatGPTInsert a sequence template as a nested clip at start_ms on a video track (default bottom video track).
list_anim_presets
ChatGPTAnimation presets (enter/exit/emphasis/procedural) with parameters. target text | video filters by clip kind.
list_branches
ChatGPTWork copies (branches) of a project with status open | requested | merged | rejected.
list_code_motions
ChatGPTCode-motion compositions in the open project (sequence id, name, prompt, param count).
list_effect_types
ChatGPTValid names for add_effect (type) and add_transition (type). Call once before using effects.
list_keyframes
ChatGPTKeyframe tracks of a clip (timeline and clip-relative times) and applied animation presets.
list_markers
ChatGPTMarkers of the active sequence.
list_pick_board
ChatGPTPicked ranges (shorts candidates) of a session.
list_projects
ChatGPTList the user's Strelizia editor projects (id, name, resolution, fps, updated_at). Use the id with open_project.
list_review_rooms
ChatGPTThe user's review rooms (shareable review links) with status, url, code flag, versions.
list_sequences
ChatGPTSequences in the open project with index, id, name, resolution, fps and which is active.
list_sessions
ChatGPTAnalysis sessions (recorded broadcasts) on this machine: id, title, date, duration, platform, chat count, whether subtitles exist. query filters by title.
list_shorts_styles
ChatGPTLayout styles for shorts: builtins (style1, none, builtin_*) and saved styles.
list_sources
ChatGPTMedia sources (video, audio, image) imported into the open project, with ids to use in add_clip.
list_templates
ChatGPTSaved sequence templates (.strz bundles) with duration and size.
list_text_presets
ChatGPTList saved text style and subtitle bar presets (id, name, kind=text_style|subtitle_bar).
list_versions
ChatGPTSaved versions of the open project, newest last, with which sequences each one touched.
move_clip
ChatGPTMove a clip to start_ms, optionally onto another track. Linked audio moves with it. One undo step.
open_code_motion
ChatGPTOpen the Code Motion window on a scene (by clip_id or sequence_id; default = selected clip or the only scene) so the user can see and edit it. new=true opens a blank window for a new scene instead.
open_project
ChatGPTOpen a project in the editor so timeline tools can act on it. Returns the project summary.
open_session
ChatGPTNavigate the app to the analysis screen and open a session (video player, chat, subtitles). Use when the user asks to open/show a session. Returns player state.
open_subtitle_designer
ChatGPTOpen the subtitle bar designer window, loaded with the current style of clip_id (default: selected clip, else the first text/caption clip).
play_session
ChatGPTPlay (playing=true, default) or pause the open analysis session's player, optionally seeking to at_ms first.
publish_review_room
ChatGPTUpload a rendered mp4 (file_path; export it first with start_export) as a new version of the open project's review room and return the guest link. Paid plan required.
recommend_hot_clips
ChatGPTRecommend the best short-clip candidates across sessions (min_sec/max_sec clip length, top_k results). 크레딧이 차감됩니다.
redo
ChatGPTRedo the last undone change.
refine_code_motion
ChatGPTEdit an existing Code Motion scene with a natural-language instruction: the built-in model rewrites the scene code (it sees the current code, params and the scene's chat history), then the scene is re-rendered in place and bound timeline clips update. Uses credits. If the edit fails, nothing changes. To write the code yourself use update_code_motion_code. One undo step.
remove_anim_preset
ChatGPTRemove one preset instance (instance_id) or all presets from a clip. Reversible with undo.
remove_effect
ChatGPTRemove one effect from a clip. Reversible with undo.
remove_keyframe
ChatGPTRemove the keyframe of a property near timeline time_ms (±16 ms).
remove_marker
ChatGPTRemove a timeline marker by id. Reversible with undo.
remove_track
ChatGPTRemove a track; force=true also deletes its clips. Reversible with undo.
remove_transition
ChatGPTRemove the transition on a clip's in or out edge. Reversible with undo.
rename_project
ChatGPTRename the open project and save.
restore_version
ChatGPTBring the project (or one sequence with sequence_id) back to a saved version. A '복원 전' version is saved automatically first.
ripple_delete
ChatGPTRemove a clip (and linked audio) and shift later clips left to close the gap. Reversible with undo.
run_auto_production
ChatGPTRun the whole automatic production pipeline on a session in the background: chat (vod_url if missing), speech subtitles, outline, highlight picks, shorts edit project. Poll get_auto_production_status; project_id appears when done. 크레딧이 차감됩니다.
run_code_motion
ChatGPTRun scene code (see get_code_motion_manual) and insert the result as a code-motion composition at start_ms (default: playhead); every node becomes real clips and keyframes. params override param() defaults. dry_run only validates and reports clip count. One undo step.
run_transcript_pipeline
ChatGPTRun subtitle post-processing on a session. stages (default repair + word_timing): filter, speaker_merge, repair, word_timing, sentence_split, speaker_match, subtitles. Background; poll get_pipeline_status. repair and subtitles: 크레딧이 차감됩니다.
save_text_preset
ChatGPTSave the current look of a caption/text clip as a named preset (a caption bar preset when a bar is attached, otherwise a text style preset). It then appears in list_text_presets and can be applied with apply_text_preset.
save_version
ChatGPTSnapshot the whole project as a named version (like a checkpoint) that restore_version can bring back.
search_transcript
ChatGPTFull-text search over all sessions' speech subtitles; returns session, time range and text. session_id narrows to one session.
seek_session
ChatGPTMove the open analysis session's player to at_ms.
set_audio_fade
ChatGPTFade-in / fade-out length in ms on a clip's audio (applied to the linked audio clip too).
set_clip_lut
ChatGPTApply a .cube LUT to a clip with intensity 0-1, or clear it (empty path). Note: one LUT texture is shared across clips.
set_clip_properties
ChatGPTChange opacity (0-1), volume (0-2), muted, speed (>0, 1 = normal), blend_mode (normal, multiply, screen, overlay, ...), transform (x/y 0-1 relative, scale_x/scale_y, rotation deg). Only given fields change. Volume applies to the linked audio too. One undo step.
set_clip_speed
ChatGPTPlayback rate 0.1-10 (2 = double speed, half length), reverse, preserve_pitch, optional duration_ms (fit), ripple to shift following clips.
set_code_motion_params
ChatGPTChange an existing Code Motion scene without rewriting its code: param values (same as dragging the sliders; values are written back into the code defaults), and/or duration_frames, speed, background, animate. The scene is re-rendered and every timeline clip bound to it updates. Call get_code_motion first to see param names, types and ranges; numbers are clamped to the declared range. One undo step.
set_keyframe
ChatGPTAdd or replace a keyframe on a clip property (position.x, position.y, scale.x, scale.y, rotation, opacity, anchor.x, anchor.y, volume, effect.<id>.<param>) at timeline time_ms (default playhead).
set_layer_board
ChatGPTOpen (open=true) or close (open=false) the layer board in a Shorts/Talk Scene project. Closing restores the timeline and side panels the way the header [Close layer board] button does.
set_review_entry_code
ChatGPTRequire (or drop) a 6-character entry code on a review room's guest link; returns the link and code.
set_shorts_backdrop
ChatGPTBackdrop color (#RRGGBB or #RRGGBBAA) behind the main video in a shorts sequence.
set_shorts_layout
ChatGPTLayout mode of the active shorts sequence: style1 | basic | facecam | facecam_off. style1 adds title, name badge and backdrop.
set_shorts_subtitles
ChatGPTTurn auto subtitles on/off on the active shorts sequence. source plain | refined; language to show translated subtitles.
set_shorts_title
ChatGPTTurn the title layer on/off and set its text.
set_shorts_video_rect
ChatGPTPosition (x, y normalized 0-1) and scale of the main video in a shorts sequence.
set_subtitle_bar
ChatGPTApply an edited caption bar / text style document (the payload from get_subtitle_bar, modified) to the clip's whole track, or to that clip only with only_clip. A layered payload (version>=3) replaces the bar: the body layer styles the captions, other layers become decorations that follow every caption. A flat payload (version 2) replaces text style and layer-style effects only. remove=true detaches the bar decorations and keeps the text style. Keep layer ids and keys you do not change. One undo step (undo also restores the previous bar document).
set_talk_config
ChatGPTChange Talk Scene settings. Per speaker (needs speaker): sync_offset_ms shifts image switching/ring timing against audio (+ later, - earlier); pad_ms keeps the speaking image longer around speech; min_ms is the minimum state length (shorter silences are bridged; default 300); visible=false removes the speaker's visuals but keeps audio, true rebuilds them; own_audio and captions toggle the speaker's own audio clips and speech captions. Scene level: background_image sets the background (empty string removes it). Give only the fields to change. Several fields in one call create several undo steps.
set_talk_image
ChatGPTReplace a Talk Scene speaker's state image (state=idle|speaking|join|leave) with an image file and re-lay that speaker's clips. speaker = display name, matched person name, or member id. The image is saved on the person profile, so later talk scenes reuse it. image_path must be a local file path (attached images are given as paths in the user message).
set_talk_speaker_outline
ChatGPTSet a speaker's frame (mask shape and diameter), frame background fill and per-state outline rings (Discord-style speaking ring). Stored in the matched person's profile (a person is created if the speaker has none), so it follows that person in other Talk Scenes. Rings and fill are drawn only when frame is circle or rounded. Give only the fields to change.
set_talk_speaker_person
ChatGPTMatch a Talk Scene speaker to a person profile (its images, animations and outline) and re-lay that speaker's clips. Empty person = unmatch (placeholder disc). Persons are listed by get_talk_scene.
set_talk_speaker_transform
ChatGPTMove and/or scale a speaker on the canvas, like dragging it in the monitor: frame mask, rings, fill and baked animation follow, and scaling also scales the frame diameter. x/y are the image center in normalized canvas coordinates (0..1, 0.5 = center); scale is the absolute image scale (current values come from get_talk_scene). Without state all of the speaker's clips move together; with state (idle, speaking or custom:<key>) only clips showing that state's image move. One undo step.
set_text_style
ChatGPTPatch text style fields on many captions at once: whole tracks (all caption clips on the track of each clip_ids entry, on track_ids, or on every text track with all_text_tracks) or only the given clips with only_clip. Only the fields you pass change. For one clip's text content use update_text; for a drop shadow use add_effect type dropShadow. One undo step.
set_timeline_panel
ChatGPTShow/hide the editor timeline panel (visible) or set its height in px (height, 40-500). Use to 'raise' or enlarge the timeline.
set_track_matte
ChatGPTUse another clip as alpha/luma matte: mode none | alpha-matte | alpha-inverted | luma-matte | luma-inverted.
set_track_state
ChatGPTMute, hide, lock, solo, volume (0-2) or name of a track.
set_volume_keyframes
ChatGPTReplace the volume automation of a clip with points [{time_ms (timeline), volume 0-2}].
split_clip
ChatGPTCut a clip (and its linked audio) at at_ms into two clips. Returns the new clip ids. One undo step.
start_chat_collect
ChatGPTCollect chat replay for a session from a VOD URL (CHZZK, YouTube, SOOP, Cime; Twitch not supported) in the background and save it. Poll get_chat_collect_status.
start_export
ChatGPTRender the active sequence to output_path (absolute; the file is overwritten). codec default h264; container/width/height/fps/bitrate_kbps optional. Returns immediately; poll get_export_status.
start_transcription
ChatGPTStart speech subtitles for a session in the background. quality: standard (default) | fast | precise. Returns immediately; poll get_transcription_status. replace=true redoes existing subtitles. 크레딧이 차감됩니다.
switch_sequence
ChatGPTMake the sequence at index the active one (timeline tools act on the active sequence).
trim_clip
ChatGPTSet a clip's new start_ms and/or end_ms on the timeline (trims the head/tail, source offset follows). Linked audio with the same edge is trimmed too. One undo step.
undo
ChatGPTUndo the last change in the editor (each write tool call is one step).
update_code_motion_code
ChatGPTReplace the full code of an existing Code Motion scene and re-run it in place; timeline clips bound to the scene update (run_code_motion always inserts a NEW scene instead). Param values with unchanged names are kept. If the code fails to run, nothing changes and the error (with line) is returned. Read get_code_motion_manual for the scene API. One undo step.
update_effect
ChatGPTMerge params into an existing effect on a clip and/or toggle enabled. One undo step.
update_pick
ChatGPTChange a pick board entry's title, comment, range, kind or source.
update_shorts_layer
ChatGPTChange one layer (clip_id) or many layers at once (clip_ids, same values to all, one undo step): text, binding key, name, hidden, position x/y (0-1), scale_x/scale_y, z_index.
update_text
ChatGPTChange a text clip's content and style: text, font_family, font_size (px), font_weight, color (#RRGGBB), align (left|center|right), bold, italic, underline, stroke_color, stroke_width (0 removes the stroke). Only given fields change. One undo step.
whoami
ChatGPTLogin state, plan, remaining credits, app version, and which project is open.