AutoClip is an open-source AI video clipping tool for creators, podcasters, and course makers who want to paste one long-video link and get platform-ready short clips with subtitles, covers, and publish copy.

What it is

AutoClip is a Python desktop application, released under the MIT licence, that cuts long-form video — podcasts, interviews, courses — into short vertical or horizontal clips. You paste a link, choose a target platform, and it produces video, a cover, and publish copy in one pass; the same one-click pipeline is shared by the desktop app, the CLI, the MCP interface, and a Docker-based web service. Editing, framing, and rendering all happen on your own machine, while the software itself is free and the cloud models it calls are billed by usage.

The concrete problem it solves is the manual repurposing workload of turning a two-hour interview into a set of platform-compliant shorts. Instead of hand-cutting segments, checking each platform's aspect ratio and duration rules, and writing captions separately, AutoClip generates up to ten qualifying clips per platform — scored, with the remainder kept as backups you can generate individually — and packages each one with a title, description, hashtags, and a ZIP publish bundle. It replaces that manual edit-and-caption workflow for Douyin, Xiaohongshu, TikTok, Instagram Reels, YouTube Shorts, Bilibili, and YouTube.

Key capabilities

• Paste a link, pick a platform, and receive up to ten scored clips plus backups, each with a cover, title, description, hashtags, and a ZIP publish package.
• Platform-specific layouts: interview-style by default for Douyin and Xiaohongshu, full-screen podcast-style for TikTok, Reels, and Shorts, and horizontal original-framing output for Bilibili and YouTube; vertical output can use either layout.
• Framing follows the active speaker, and shots with nobody on screen keep the full frame; an automatic brand outro is enabled by default and can be switched off in settings.
• Subtitles and publish copy follow the platform's language, with subtitle pagination among the fixes shipped in v1.5.0.
• Transcription is chosen per project: author subtitles (which skip transcription entirely), local Whisper or SenseVoice, or a configured cloud transcription service.
• Analysis models are selectable, from cloud services such as qwen-plus to local models served through Ollama or LM Studio.
• Distribution of the tool itself spans desktop installers (.dmg for macOS Apple Silicon, -setup.exe for Windows 10/11 x64, both bundling Python and FFmpeg), CLI/MCP, and a Docker Compose web deployment.

Who uses it and how

• Desktop users on macOS Apple Silicon or Windows x64 install from Releases, configure a provider and API key in settings, paste a link, preview the generated clips, and download the publish package or connect an account to submit.
• Self-hosters run it as a web service with Docker and Compose v2 on Linux, exposing the web interface at localhost:3000 and API documentation at localhost:8000/docs.
• Batch and agent workflows use the CLI/MCP path with Python 3.10+ (3.11 recommended) and FFmpeg, requiring separate setup from the bundled desktop installers.
• Measured in practice on an Apple Silicon Mac on 2026-10-01: a 1h43m English interview to Xiaohongshu produced 10 clips in 7.5 minutes at an estimated ¥0.09 in qwen-plus text-model cost, while a 2h52m clipless Chinese source to Douyin took 29.5 minutes at roughly ¥0.20.

Getting started

Download the desktop build from the Releases page, or clone the repository and run docker compose up -d --build after copying env.example to .env to reach the web interface.

How it compares

AutoClip stands alone in this registry; no comparable tools are named in its documentation.

When to use it — and when not

A self-hoster must operate the Docker stack with its data, logs, and uploads directories, or supply Python 3.10+ and FFmpeg for the CLI, and must configure either cloud API keys or local Whisper/SenseVoice and Ollama/LM Studio services — cloud analysis sends subtitles and copy, cloud transcription sends audio, and image understanding sends sampled frames. It is a poor fit for anyone unwilling to configure a model provider, for Intel Mac and Linux users who would need Docker or the CLI rather than a desktop installer, and it carries ongoing cost because the free software leans on pay-per-use cloud models, with 13 issues currently open.

git clone https://github.com/zhouxiaoka/autoclip.git
cd autoclip
cp env.example .env
mkdir -p data logs uploads
docker compose up -d --build

docker compose run --rm --no-deps --user root --entrypoint sh autoclip -c 'chown -R autoclip:autoclip /app/data /app/logs /app/uploads'
docker compose up -d

python3 -m venv venv
source venv/bin/activate
python -m pip install -r requirements.txt
python -m pip install --force-reinstall --no-deps autoclip-1.5.0-py3-none-any.whl

autoclip --version
autoclip produce talk.mp4 --srt talk.srt --platform douyin --portrait-style podcast --json
autoclip outputs PROJECT_ID --export-kits