agricidaniel/claude-shorts — explained in plain English
Analysis updated 2026-05-18
Turn a long podcast or talk recording into several short vertical clips ready for social media.
Automatically caption vertical video clips in different visual styles.
Extract highlight moments from screen recordings or talking-head videos using AI scoring.
| agricidaniel/claude-shorts | xzf-thu/mega-asr | paddlepaddle/graphnet | |
|---|---|---|---|
| Stars | 93 | 93 | 95 |
| Language | Python | Python | Python |
| Last pushed | — | — | 2026-05-22 |
| Maintenance | — | — | Maintained |
| Setup difficulty | moderate | hard | easy |
| Complexity | 3/5 | 4/5 | 2/5 |
| Audience | developer | researcher | researcher |
Figures from each repo's GitHub metadata at analysis time.
Needs FFmpeg, Node.js, Python, and Claude Code installed, a GPU is recommended but not required.
claude-shorts is a Claude Code skill, a plugin for Anthropic's command-line AI tool, that takes a long video file and automatically extracts short, vertical clips suitable for YouTube Shorts, TikTok, and Instagram Reels. The process is interactive: Claude guides you through each step, presents candidate clips for your approval, and then renders the final output with animated captions. The pipeline works in ten steps. First it validates your input video and checks your system. Then it transcribes the full audio using a GPU-accelerated transcription tool called faster-whisper, which produces word-level timestamps. Next, it detects what type of content is in the video, talking-head footage, screen recordings, or podcasts, because each type is handled differently during the crop-and-reframe step. Claude then reads the full transcript and scores eight to twelve candidate segments using a five-dimension rubric evaluating hook strength, narrative coherence, emotional content, information density, and payoff. You see these candidates in a table with scores and can choose which ones to render, pick a caption style, and select your target platform. The tool then aligns your cut points to natural word boundaries and silences to avoid awkward mid-word cuts, extracts the clips, converts the wide video to vertical 9:16 format, and renders animated captions using Remotion (a React-based video rendering library). Three caption styles are available: bold all-caps with yellow highlights, bouncy colorful text, and clean minimal fade-in. The final export is encoded to the exact specifications each platform prefers. For screen recordings, the cropping dynamically follows the mouse cursor. For talking-head footage, it tracks faces. GPU acceleration is used where available but the tool falls back to CPU. The full README is longer than what was provided.
A Claude Code skill that turns long videos into short vertical clips for TikTok, YouTube Shorts, and Instagram, with AI picking the best moments and adding animated captions.
Mainly Python. The stack also includes Python, Node.js, Remotion.
No license file mentioned in the README.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.