zhouxiaoka/autoclip

AutoClip cuts long-form videos such as podcasts, interviews, and courses into short clips tailored to platform requirements, supporting Douyin, Xiaohongshu (RED), TikTok, Reels, Shorts, Bilibili, and YouTube: For each platform, it automatically generates up to 10 clips based on scoring, with the rest listed as alternatives; each clip includes a cover image, title, description, hashtags, and a publishing package. Vertical formats offer two layouts—interview style and full-screen podcast style—with the camera following the speaker. Analysis models can be chosen from cloud providers or local models via Ollama and LM Studio; transcription is skipped if author-provided subtitles are available.

AutoClip cuts long-form videos such as podcasts, interviews, and courses into short clips tailored to platform requirements, supporting Douyin, Xiaohongshu (RED), TikTok, Reels, Shorts, Bilibili, and YouTube: For each platform, it automatically generates up to 10 clips based on scoring, with the rest listed as alternatives; each clip includes a cover image, title, description, hashtags, and a publishing package. Vertical formats offer two layouts—interview style and full-screen podcast style—with the camera following the speaker. Analysis models can be chosen from cloud providers or local models via Ollama and LM Studio; transcription is skipped if author-provided subtitles are available.

Why It Matters Editing, subject framing, and rendering are performed locally; only content analysis and optional cloud transcription require model calls. The desktop version, command-line interface, and MCP share a single video output pipeline.

Star count 11248 (+126 / 24h)

Reasons

  • 126 new stars in 24 hours, total stars 11248 (indicating attention only); MIT license
  • The author lists time and cost for three sample materials: A 1 hour 43 minute English interview produces 10 clips in approximately 7.5 minutes, with estimated text model costs of 0.09 yuan

Usage

  • Download the desktop version from Releases for macOS Apple Silicon or Windows, select a provider in settings, enter the API Key, and test the connection, then paste the video link and select the platform; to integrate with a proxy, run uvx autoclip-mcp

Risks

  • Most README examples are clips from celebrity interviews; the original footage copyright belongs to the creators, so ensure you have the right to use the materials before publishing
  • Cost figures are estimated by the author on their own Mac using qwen-plus, excluding costs for cloud speech recognition, image generation, and submission services

Open

Trendshift(GitHub Trending 快照)

Explore the latest AI Radar updates