Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
What This Skill Does
Pipeline that converts a script into a multi-shot video through five ordered stages: outline, episode_plan, storyboard, storyboard_images, and render. Each stage writes a checkpoint file and halts until explicitly confirmed, letting users review and correct outputs before proceeding. Uses Seedream for storyboard image generation and Seedance for video rendering, with optional FFmpeg concatenation into a final MP4.
The checkpoint confirmation system prevents wasted compute on downstream stages when earlier outputs need revision.
When to use it
- Converting a written story script into a storyboard video
- Reviewing and approving each production stage before rendering begins
- Resuming a paused video pipeline from a saved checkpoint
- Generating multi-shot video sequences from structured JSON episode plans
- Merging individually rendered video clips into a single MP4 file
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: Convert story.txt into a final video
- 1Run run_story.py with --stage render, pipeline halts at outline checkpoint
- 2User reviews outline, runs confirm --stage outline, pipeline continues
- 3User confirms episode_plan, storyboard, and storyboard_images checkpoints in sequence
- 4Seedream generates images for each storyboard shot
- 5Seedance renders video clips per shot, concat_videos.py merges them into final-video.mp4
final-video.mp4 saved to videos/run-YYYYMMDD-HHMMSS/final-video.mp4
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.