claude-shorts
CLAUDE.md203 starsChanged 8 months agoMITView on GitHub
Standalone interactive shortform video creator. Extracts viral-ready short clips from longform videos using an interactive Claude Code workflow with Remotion-rendered premium captions. Scripts use ~/.video-skill/ venv if it exists (shared with claude-video), otherwise ~/.shorts-skill/. Activate before running Python scripts: The Remotion project lives in remotion/. After bash setup.sh: Pipeline stores intermediate files in $SHORTS_TMP (defaults to /tmp/claude-shorts/). Override with export SHORTS_TMP="/path/to/dir" for systems with limited /tmp space. The directory is created automatically and cleaned between runs.
In this repositoryOne other file it gives its agents.
- shortsSkill
The file
Files
claude-shorts
Standalone interactive shortform video creator. Extracts viral-ready short clips from longform videos using an interactive Claude Code workflow with Remotion-rendered premium captions.
Architecture
- Claude Code is the orchestrator — reads transcripts, scores segments, presents to user interactively
- faster-whisper handles GPU-accelerated transcription with word-level timestamps
- Remotion renders 1080x1920 vertical video with animated captions (single-pass)
- FFmpeg extracts audio, cuts segments (stream copy), and does platform-specific export encoding
Key Commands
bash setup.sh # Install Python + Node dependencies (one-time)
bash install.sh # Copy skill to ~/.claude/skills/shorts/Python Virtual Environment
Scripts use ~/.video-skill/ venv if it exists (shared with claude-video), otherwise ~/.shorts-skill/. Activate before running Python scripts:
source ~/.video-skill/bin/activate # or ~/.shorts-skill/bin/activateRemotion
The Remotion project lives in remotion/. After bash setup.sh:
cd remotion && npx remotion preview # Preview compositions in browser
node remotion/render.mjs # Headless render (used by pipeline)Temp Files
Pipeline stores intermediate files in $SHORTS_TMP (defaults to /tmp/claude-shorts/). Override with export SHORTS_TMP="/path/to/dir" for systems with limited /tmp space. The directory is created automatically and cleaned between runs.
Dependencies
- FFmpeg (system) — audio extraction, segment cutting, export encoding
- Python 3.10+ with faster-whisper, mediapipe, numpy, opencv-python
- Node.js 18+ with remotion, @remotion/captions, zod
- NVIDIA GPU recommended (NVENC encoding, CUDA transcription)
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

