π«§ Video Edit β Pro Pack on RunComfy
by @kalvinrv
Video edit on RunComfy. This video edit skill transforms an existing video clip β restyle, background swap, outfit swap, motion transfer, color grade, or any...
clawhub install video-edit-runcomfyπ About This Skill
name: video-edit-runcomfy displayName: "π«§ Video Edit β Pro Pack on RunComfy" description: > Video edit on RunComfy. This video edit skill transforms an existing video clip β restyle, background swap, outfit swap, motion transfer, color grade, or any other video edit task β by routing the video edit request to the right model in the RunComfy catalog. Video edit supports talking-head video edit, product video edit, and short-form video edit at up to 1080p. Calls
runcomfy run /edit-video
through the local RunComfy CLI. Triggers on "video edit", "edit
video", "video editing", "video-edit", "restyle video", "swap video
background", "video outfit swap", "video color grade", or any
explicit ask to edit a video.
emoji: "π«§"
homepage: https://www.runcomfy.com
license: MIT
clawdis:
requires:
bins:
- runcomfy
env:
- RUNCOMFY_TOKEN
config:
- ~/.config/runcomfy
π«§ Video Edit β Pro Pack on RunComfy
runcomfy.com Β· docs Β· Video edit models
Video edit on RunComfy. This skill is the canonical video edit entry point for the RunComfy Model API: give it a source video URL and an edit instruction, and it returns the edited video. Video edit on RunComfy means transforming an existing clip β restyle, background swap, outfit swap, motion transfer, color grade β without re-shooting.
What "video edit" means here
Video edit is the task of taking a source video and producing a transformed video that preserves identity, motion, or framing where you want, while changing what you specify. Video edit is distinct from text-to-video (no input clip) and from image-to-video (input is a still). Common video edit operations include:
This skill picks the right video edit endpoint for the user's intent and calls runcomfy run with the matching schema.
When to use video edit on RunComfy
Pick video edit on RunComfy whenever:
If the user said "video edit", "edit video", "restyle this video", "swap the background", "change the outfit", "transfer this motion", "color grade this clip", or showed a video and asked to transform it β route here.
Video edit routes
| User intent | Video edit model | Why |
|---|---|---|
| Default video edit β restyle, background swap, color grade, packaging swap | wan-ai/wan-2-7/edit-video | Most versatile video edit model; identity + motion preservation, up to 1080p video edit output |
| Motion-transfer video edit (transfer motion from a reference clip) | kling/kling-2-6/motion-control-pro | Designed for motion-mapping video edit with identity hold |
| Lightweight outfit-swap / atmospheric restyle video edit | decart/lucy-edit/restyle | Fastest video edit pass for localized style changes; 720p |
The agent reads this table, classifies the user's video edit intent, and picks the matching endpoint.
Prerequisites
1. RunComfy CLI β npm i -g @runcomfy/cli
2. RunComfy account β runcomfy login.
3. CI / containers β set RUNCOMFY_TOKEN=.
4. A source video URL β formats and limits depend on the chosen video edit route.
Default video edit β Wan 2.7 Edit-Video
The default video edit endpoint. Use for any general video edit task: restyle a talking-head video, swap a product background, replace packaging design with a reference image, apply a cinematic color grade. Up to 1080p video edit output.
Schema
| Field | Type | Required | Default | Notes |
|---|---|---|---|---|
| prompt | string | yes | β | Video edit instruction. Lead with preservation goals, then state the change. |
| video | string | yes | β | Source video URL for video edit. MP4/MOV, 2β10s, β€100MB. |
| reference_image | string | no | β | Optional reference for design-transfer video edit (e.g. packaging swap). |
| resolution | enum | no | (input) | 720p or 1080p for the video edit output. |
| aspect_ratio | enum | no | (input) | W:H. Defaults to source video aspect. |
| duration | int | no | 0 | 0 = match input; 2β10 truncates the video edit from the start. |
| audio_setting | enum | no | auto | auto regenerates audio; origin preserves source audio in the video edit output. |
| seed | int | no | β | Reproducibility for video edit variants. |
Invoke
Background swap video edit, identity preserved, audio kept:
runcomfy run wan-ai/wan-2-7/edit-video \
--input '{
"prompt": "Preserve the speaker'\''s face, pose, and lip movement; change the background to a modern office with neutral lighting.",
"video": "https://.../speaker.mp4",
"audio_setting": "origin"
}' \
--output-dir
Packaging-swap video edit with reference image:
runcomfy run wan-ai/wan-2-7/edit-video \
--input '{
"prompt": "Maintain the original framing and hand movement; replace the packaging design using the reference image.",
"video": "https://.../hand-holding-package.mp4",
"reference_image": "https://.../new-packaging.png",
"audio_setting": "origin"
}' \
--output-dir
Motion-transfer video edit β Kling 2.6 Pro Motion Control
Use when the video edit transfers motion from a reference clip onto a target character. This isn't restyle video edit β it's motion-mapping video edit with identity hold.
| Field | Type | Required | Notes |
|---|---|---|---|
| prompt | string | yes | Describe the target motion / style for the video edit output. |
| image | string | yes (image orientation) | Reference for character / background consistency in the video edit. |
| video | string | yes | Motion-reference clip for the video edit. 10β30s depending on orientation. |
| keep_original_sound | bool | no | Preserve audio from the reference video edit input. |
| character_orientation | enum | yes | image (max 10s video edit output) or video (max 30s). |
runcomfy run kling/kling-2-6/motion-control-pro \
--input '{
"prompt": "A young american woman dancing",
"image": "https://.../target-character.jpg",
"video": "https://.../motion-reference-dance.mp4",
"character_orientation": "image",
"keep_original_sound": true
}' \
--output-dir
Lightweight video edit β Lucy Edit Restyle
Use when the video edit is a localized style modification β outfit swap, scene relight, atmospheric restyle β and identity preservation is critical. Faster and cheaper than Wan 2.7 Edit-Video; capped at 720p.
| Field | Type | Required | Default | Notes |
|---|---|---|---|---|
| prompt | string | yes | β | Natural-language video edit instruction. |
| video_url | string | yes | β | MP4/MOV/WEBM/GIF source for the video edit. |
| resolution | enum | no | 720p | 720p only for this video edit tier. |
Outfit-swap video edit:
runcomfy run decart/lucy-edit/restyle \
--input '{
"prompt": "Change outfit to professional business attire; preserve face and motion.",
"video_url": "https://.../subject-walking.mp4"
}' \
--output-dir
Atmospheric video edit:
runcomfy run decart/lucy-edit/restyle \
--input '{
"prompt": "Make lighting warm and golden hour; preserve face, pose, and motion.",
"video_url": "https://.../subject-portrait.mp4"
}' \
--output-dir
Prompting video edit β what works
Video edit prompts behave differently from text-to-video prompts. The source clip already fixes most of the look β your prompt should drive the change, not redescribe the video.
"Preserve [face / pose / motion / framing / lip movement]; [then state the video edit change]". Tell the video edit model what NOT to change.reference_image only when the video edit needs an exact visual (packaging swap, costume swap matching a target). Don't pass refs for general restyle video edit.audio_setting: "origin" for talking-head video edit where you don't want the soundtrack regenerated.Video edit FAQ
What's the max duration of a video edit clip? Wan 2.7 Edit-Video: 2β10s. Kling Motion Control: 10s (image orientation) or 30s (video orientation). Lucy Edit Restyle: matches input.
What video formats does video edit accept? MP4, MOV (Lucy also takes WEBM and GIF). Source video edit input must be β€100MB on Wan 2.7.
Does video edit preserve face identity? Yes β all three video edit routes are designed for identity preservation. State the goal explicitly: "preserve face and motion".
Can video edit keep the original audio? Yes β set audio_setting: "origin" on Wan 2.7 Edit-Video, or keep_original_sound: true on Kling. Lucy preserves audio by default.
What's the highest-resolution video edit available here? 1080p on Wan 2.7 Edit-Video. Kling and Lucy cap at 720p.
Video edit vs text-to-video on RunComfy? Video edit transforms an existing clip (look largely fixed by source). Text-to-video starts from a prompt only (look generated). Use video edit when you have a clip; use text-to-video for novel content.
Can I run multiple video edits in one call? No. Each video edit call applies one direction; for compound video edits, chain calls and stitch.
Limitations
wan-2-7) instead of forcing it through this video edit router.Exit codes
| code | meaning | |---|---| | 0 | video edit succeeded | | 64 | bad CLI args | | 65 | bad input JSON for video edit / schema mismatch | | 69 | upstream 5xx | | 75 | retryable: timeout / 429 | | 77 | not signed in or token rejected |
Full reference: docs.runcomfy.com/cli/troubleshooting.
How it works
The skill picks one of three video edit endpoints (Wan 2.7 Edit-Video, Kling Motion Control, or Lucy Edit Restyle) based on user intent, and invokes runcomfy run with the matching JSON body. The CLI POSTs to the RunComfy Model API, polls the video edit request status every 2 seconds, and downloads the resulting video from the *.runcomfy.net / *.runcomfy.com URL into --output-dir. Ctrl-C cancels the in-flight video edit request.
Security & Privacy
runcomfy login writes the API token to ~/.config/runcomfy/token.json with mode 0600. Set RUNCOMFY_TOKEN env var in CI.--input. The CLI does NOT shell-expand. No shell-injection surface.model-api.runcomfy.net and *.runcomfy.net / *.runcomfy.com. No telemetry.βοΈ Configuration
1. RunComfy CLI β npm i -g @runcomfy/cli
2. RunComfy account β runcomfy login.
3. CI / containers β set RUNCOMFY_TOKEN=.
4. A source video URL β formats and limits depend on the chosen video edit route.