Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@Arkeonix CapCut MCP Align subtitles to narration in the current draft."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
An MCP server for inspecting and editing CapCut Desktop draft projects through local, version-safe timeline operations. It is designed for AI-assisted documentary, social, and narration-first editing workflows.
The server edits draft files locally. It does not automate the CapCut UI or upload media to a service.
Read, clone, validate, save, and restore local CapCut drafts.
Place and transform images, video, audio, subtitles, text, stickers, and timeline effects.
Align ordered narration, visual, and subtitle clips from the real audio duration.
Create keyframe motion, typography overlays, transitions, filters, masks, audio fades, and B-roll layers.
Inspect timeline health before saving, with backups and protection against writing while CapCut is open.
Related MCP server: capcut-mcp
Node.js 18 or newer.
FFmpeg and FFprobe available on PATH.
CapCut Desktop.
On macOS, install FFmpeg with brew install ffmpeg. On Windows, use
winget install Gyan.FFmpeg.
git clone https://github.com/daviilpzDev/arkeonix-capcut-mcp.git
cd arkeonix-capcut-mcp
npm install
npm testThe optional setup script also checks prerequisites, installs dependencies, syncs locally available application assets, and installs a clean draft template:
./setup.shDownloaded application assets are intentionally ignored by Git. They can be large and may be governed by separate vendor terms.
Use a local MCP entry such as:
{
"mcp": {
"arkeonix-capcut": {
"type": "local",
"command": [
"node",
"/ABSOLUTE/PATH/TO/arkeonix-capcut-mcp/src/server.js"
],
"enabled": true,
"environment": {
"CAPCUT_DRAFTS_DIR": "",
"CAPCUT_TEMPLATE_DRAFT": ""
}
}
}
}CAPCUT_DRAFTS_DIR is optional when CapCut uses its standard location.
CAPCUT_TEMPLATE_DRAFT optionally names a draft used as the material template
when a target draft lacks a required media type.
The server also supports optional vision and B-roll integrations through
GEMINI_API_KEY, OPENROUTER_API_KEY, and MAGNIFIC_API_KEY. Keep these in
your local environment or MCP configuration, never in the repository.
Read the timeline and validate before making editorial changes.
Clone a production draft before modifying it.
Do not save while CapCut is running. The server refuses this by default.
Saving writes a .mcpbak backup and validates the resulting draft.
After saving, reload and inspect the draft to verify persistence.
For narration-first projects, make the narration duration the timing source. Align ordered audio and subtitle clips, preserve every source duration, then align visuals. For emphasis overlays, retain base subtitles but set them invisible, place exact phrases at their spoken position, and use short, non-overlapping word-reveal segments on a separate text track.
The included MCP tools use the capcut_ prefix to make the target draft format explicit.
Arkeonix CapCut MCP is a maintained distribution of the MIT-licensed
minminthaw/capcut-mcp project,
with Arkeonix workflow, reliability, and packaging changes. The original MIT
copyright notice is preserved in LICENSE.
Add an audio clip at a time on a track.
Inputs: file (required), atSec (required), draft (required), posX, +10 more.
Add modern dynamic word-highlight subtitles (Alex Hormozi style) with pop colors and heavy outline.
Inputs: draft (required), subtitles (required), posY, fontSize, +4 more.
Add an image at a time on a track.
Inputs: file (required), atSec (required), draft (required), posX, +10 more.
Add smooth animation keyframes (scale, position, rotation, opacity, volume) to a segment.
Inputs: draft (required), property (required), keyframes (required), segmentId (required).
Add a speaker name and title badge / lower-third overlay with animations.
Inputs: draft (required), title (required), posX, posY, +6 more.
Add an animated video progress bar that fills from 0% to 100% across the video duration.
Inputs: draft (required), color, position, trackIndex.
Place a sticker/overlay onto the timeline at a specific time, position, scale, and duration.
Inputs: atSec (required), draft (required), sticker (required), posX, +6 more.
Bulk add Burmese/English subtitles with typography styling, outline strokes, and positions.
Inputs: draft (required), subtitles (required), posY, color, +4 more.
Add a text overlay. Requires a text template (a draft with a text layer; see CAPCUT_TEMPLATE_DRAFT).
Inputs: text (required), atSec (required), draft (required), posX, +5 more.
Add a new track (video | audio | text | sticker | effect | filter).
Inputs: draft (required), name, type.
Add a video clip at a time on a track. Session edit; call capcut_save to persist.
Inputs: file (required), atSec (required), draft (required), posX, +10 more.
Align audio, image and subtitle segments by current order, using each audio duration.
Inputs: draft (required), textTrackIndex, videoTrackIndex.
Extract keyframes from video and analyze scene composition, emotion, visual actions, and recommended CapCut edits using Multimodal AI.
Inputs: draft (required), videoPath (required), model, apiKey, +3 more.
Apply an in/out/loop animation to a video clip, image, or text overlay.
Inputs: draft (required), animation (required), segmentId (required), startSec, +2 more.
Apply a voice filter or audio scene effect (e.g. "Robot", "Deep", "Echo", "Telephone", "Chipmunk").
Inputs: draft (required), segmentId (required), audioEffect (required).
Apply Green Screen / Chroma Key background cutout with intensity and shadow.
Inputs: draft (required), segmentId (required), color, shadow, +1 more.
Apply a visual effect to a specific segment or as a standalone effect layer track.
Inputs: draft (required), effect (required), params, startSec, +4 more.
Apply a color filter (grade/look) to a specific segment or as a standalone adjustment filter track.
Inputs: draft (required), filter (required), startSec, intensity, +3 more.
Apply a video shape mask (circle, rectangle, linear, mirror, heart, star) with feather and inversion.
Inputs: draft (required), maskType (required), segmentId (required), width, +7 more.
Apply Picture-in-Picture or Split-Screen layout preset to a clip.
Inputs: draft (required), layout (required), segmentId (required), scale.
Apply a transition after the selected clip segment (transitioning into the following clip).
Inputs: draft (required), segmentId (required), transition (required), overlap, +1 more.
Automatically duck/lower BGM music volume during spoken dialogue and raise volume during silence.
Inputs: draft (required), bgmTrackIndex (required), baseVolume, duckVolume, +2 more.
Automatically place B-Roll cutaways (videos/images) on an overlay track at designated timestamps.
Inputs: draft (required), insertions (required), trackIndex.
Automatically slice and ripple-delete silence dead space on the main video track based on speech intervals.
Inputs: draft (required), speechIntervals (required), paddingSec, trackIndex.
Copy a draft folder to a new name (valid scaffolding). empty:true clears all clips/tracks for a fresh build.
Inputs: base (required), newName (required), empty.
Remove a segment.
Inputs: draft (required), segmentId (required).
Drop unsaved session edits and reload the draft from disk.
Inputs: draft (required).
Search and retrieve timestamped scenes from the video understanding map by visual query or emotion (e.g. "crying", "phone demo", "high energy").
Inputs: draft (required), query (required).
Generate structured YouTube / Social Media chapter markers and timestamps from scene analysis.
Inputs: draft (required).
Analyze and inspect timeline statistics, B-Roll coverage percentage, cut count, and validation health.
Inputs: draft (required).
Search CapCut animations catalog (~800+ entries: in, out, loop, group for video or text).
Inputs: type, limit, query, cachedOnly, +1 more.
List CapCut drafts in /root/Movies/CapCut/User Data/Projects/com.lveditor.draft with duration and lock status.
Inputs: no named parameters.
Search CapCut visual/character effects catalog (~1800+ entries) by keyword or category.
Inputs: limit, query, category, cachedOnly.
Search CapCut color filters catalog (~450+ LUTs/looks, e.g. "Vintage 90s", "American Film", "Teal & Orange").
Inputs: limit, query, cachedOnly.
Search available and cached CapCut stickers, emojis, subscribe buttons, and badges.
Inputs: limit, query.
Search CapCut transitions catalog (~1100+ entries, e.g. "Cross Dissolve", "Bubble Blur", "Flash White").
Inputs: limit, query, cachedOnly.
Generate or upscale a cinematic 4K/8K B-Roll stock visual using Magnific AI API and insert into timeline with Ken Burns slow zoom.
Inputs: draft (required), prompt (required), startSec (required), hdr, +7 more.
Move a segment to a new start time and optionally another track.
Inputs: atSec (required), draft (required), segmentId (required), trackIndex.
Normalize dialogue/voice volume across all segments on a track.
Inputs: draft (required), trackIndex, targetVolume.
Advanced escape hatch: deep-merge a JSON patch into draft_content (undocumented ops). Use with care.
Inputs: draft (required), patch (required).
Read a draft: canvas, fps, tracks and every segment (id, media, times, layer). Read-only.
Inputs: draft (required).
Write session edits to disk (backs up .mcpbak, validates). Refuses if CapCut is open unless force:true.
Inputs: draft (required), force.
Automatically apply a CapCut editing action (filter, effect, zoom, lower_third, canvas_blur) matching a visual scene query.
Inputs: draft (required), query (required), action (required), name, +4 more.
Set fade-in and fade-out durations on an audio or video clip.
Inputs: draft (required), segmentId (required), fadeInSec, fadeOutSec.
Change project canvas aspect ratio (e.g. "9:16" for TikTok/Reels, "16:9" for YouTube, "1:1" for Square).
Inputs: draft (required), ratio (required), width, height.
Apply Gaussian canvas background blur or solid color behind a clip for vertical/horizontal reframing.
Inputs: draft (required), segmentId (required), type, color, +1 more.
Set manual slider color adjustments (brightness, contrast, saturation) on a segment.
Inputs: draft (required), segmentId (required), contrast, brightness, +1 more.
Set transform / opacity / volume / speed / visibility on a segment.
Inputs: draft (required), segmentId (required), posX, posY, +8 more.
Apply dynamic speed ramping/curves (e.g. "montage", "hero", "bullet_time", "flash_in", "flash_out") to a video clip.
Inputs: curve (required), draft (required), segmentId (required).
Split a segment into two at a timeline time.
Inputs: atSec (required), draft (required), segmentId (required).
Synchronize timeline edits (cuts or punchy zoom pulses) to music beat markers.
Inputs: draft (required), beatTimestamps (required), action, zoomScale.
Change a segment start / duration / source in-point (seconds).
Inputs: draft (required), segmentId (required), atSec, durSec, +1 more.
Check the (in-session) draft for overlaps, duplicate ids/render_index, dangling references, missing media.
Inputs: draft (required).
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 53 tools
Most tools target clearly distinct resources (e.g., list_effects vs apply_effect, add_video vs add_text). A few pairs could cause misselection, such as capcut_add_subtitles_batch vs capcut_add_dynamic_captions or capcut_apply_filter vs capcut_set_color_adjustments, but descriptions generally resolve ambiguity.
Tools consistently use the capcut_ prefix followed by verb_noun patterns like list_, apply_, add_, set_, and auto_. Minor deviations include capcut_raw_patch and capcut_magnific_enhance_broll, but overall the convention is predictable and readable.
With 53 tools, the surface is very large and exceeds the comfortable range for agent navigation. While the video-editing domain is broad, many specialized one-off tools (e.g., capcut_add_progress_bar, capcut_generate_chapters) could be consolidated or composed from simpler primitives.
The tool set covers the full editing lifecycle: reading drafts, adding/removing segments, transforming clips, applying effects and transitions, managing audio, and even advanced AI-driven scene analysis and auto-editing. No major gaps are apparent for working with CapCut draft files.
A MCP server built for developers enabling Git based project management with project and personal…
MCP server for Hailuo (MiniMax) AI video generation
MCP server for generating rough-draft project plans from natural-language prompts.
MCP server for developers documentation, generated by doc2mcp.
