AutoShorts is an open-source, local-first desktop app published on GitHub that turns long videos and podcasts into ranked, vertical short-form clips for Shorts, Reels and TikTok. The project has gathered around 1,300 GitHub stars and is actively developed in 2026.
Built with Tauri, React and Rust, AutoShorts runs the whole clipping pipeline on your machine: it imports a long recording, extracts the audio, transcribes it, then uses an LLM to detect and rank the moments most likely to perform as standalone shorts, before centre-cropping landscape footage to 9:16 portrait with ffmpeg and burning captions. You can work fully offline with local Ollama models plus Whisper transcription, or plug in cloud APIs โ DeepSeek or Claude for moment detection and hooks analysis, Deepgram for transcription โ with per-run costs of a fraction of a cent on the recommended setup. Projects, transcripts and candidates live in a local SQLite database managed from a native dashboard, and installers are provided for macOS, Windows and Linux. Nothing is uploaded to a vendor’s cloud unless you choose a cloud engine.
AutoShorts is free and open source. Offline use costs nothing; optional cloud APIs bill to your own accounts at roughly $0.001โ0.05 per run.
Podcasters, streamers and creators who repurpose long-form content into shorts and want the workflow on their own computer instead of a subscription web service.