Route natural-language video editing intents to RunComfy CLI model calls.
npx skills add https://github.com/runcomfy-com/skills --skill video-edit-runcomfy-com
Please help me install this Agent Skill. Skill: video-edit Source: https://github.com/runcomfy-com/skills/tree/main/video-edit Command: npx skills add https://github.com/runcomfy-com/skills --skill video-edit-runcomfy-com
Video editing from natural-language intent is slow and error-prone because selecting the right underlying model and prompting pattern is difficult, especially when the goal is restyling versus motion transfer.
video vs video_url, reference_image, and motion-control orientation constraints.runcomfy run <vendor>/<model>/<endpoint> call and downloads results to your output directory.Use case: You want to keep a talking-head’s face and lip movement while swapping the background to a modern office and preserving the original audio; the skill routes you to the best-fit edit model and applies the appropriate prompting structure.
Ask the agent to edit your video by preserving face and motion while changing the background to a modern office, and include your source video URL.
High-intent search queries and answers about installing and using this skill.
Identity preservation in video restyling maintains a talking-head's face and lip movement while changing the background or outfit. The skill routes your request to a lightweight restyle model that enforces identity-stable constraints across character-edit scenarios.
Model-specific input handling for video editing enforces valid schema fields like video versus video_url and reference_image based on the routed model. It ensures your JSON inputs match the exact requirements of the selected restyle or motion-control endpoint.
Yes, you can use natural language to request a video edit, such as preserving face and motion while changing the background to a modern office. The skill parses your intent, routes it to the best-fit edit model, and applies the appropriate prompting structure automatically.
Automated intent routing for video edits is limited to restyle, background packaging, motion transfer, and identity-stable outfit changes. It enforces model-specific constraints, meaning unsupported edit types or invalid schema fields will fail the RunComfy CLI invocation.
prime-skillsRoute video editing requests to RunComfy models and execute transformations.
OpenLoafProcess local media files and download public videos for editing.
runcomfy-comRoute face swaps in images and videos to RunComfy model endpoints.
agentspace-soRoute video editing intents to RunComfy models and execute endpoints.