OpenClip
A skill that pulls highlights from a long video: download, transcript, segmentation, analysis and clips with titles
Medium risk
We rate an entry medium when the tool runs code, makes network calls or reads project files. Check what exactly it does before installing.
Why this level
- Downloads video and runs ffmpeg and a Python pipeline
- Reaches an LLM provider over the network with a key
Install
Manual install
npx skills add https://github.com/linzzzzzz/openclip --skill video-clip-extractorInstalls the video-clip-extractor skill. Requires ffmpeg and uv.
This is third-party code. Review the repository files before installing.
What it does
The skill runs a pipeline that finds and cuts the best moments from long videos and stream recordings. From a URL or a local file it downloads the video, transcribes it, splits the transcript, analyzes the content with a model and assembles clips with titles and covers. It offers clip-length presets, a deep boundary-optimization mode, subtitle burn-in with translation and a user-intent hint to prioritize a topic. It works with Qwen, GLM, MiniMax, OpenRouter and a custom OpenAI-compatible endpoint.
Who it is for. For creators, editors and social media managers who need short clips from long videos and streams.
Good fit when
- You need highlights cut from a long video or a stream recording
- You need short clips with titles and covers for social media
- You want fragments on a specific topic via user-intent
Not a fit when
- You have no ffmpeg and cannot set up the Python environment via uv
- You have no key for any supported LLM provider
Example request
Cut short clips with the best moments and titles from this stream recordingLimitations
It needs ffmpeg and a Python environment via uv, and ffmpeg with libass for subtitle burn-in. It needs a key for at least one LLM provider: Qwen, OpenRouter, GLM or MiniMax. Sources are Bilibili and YouTube: some sites and providers may need a VPN from Russia, while local files and Chinese providers work directly. Speaker identification is in preview and needs extra setup and a Hugging Face token.
How to disable. Remove the video-clip-extractor skill folder from the agent's skills directory and the openclip install at ~/.local/share/openclip.
Security check
- Downloads video and runs ffmpeg and a Python pipeline
- Reaches an LLM provider over the network with a key
README in short
The README describes a lightweight video pipeline: download, transcript, segmentation, model analysis and clips with titles and covers. It ships a Streamlit web UI, a CLI and an Agent Skill install via the skills CLI. Sources are Bilibili, YouTube and local files, and providers include Qwen, GLM, MiniMax, OpenRouter and a custom OpenAI-compatible endpoint. Features include length presets, deep boundary optimization, subtitle burn-in and a user-intent hint. MIT licensed.
FAQ
Which LLM provider do I need?
One key is enough: Qwen by default, or OpenRouter, GLM or MiniMax. You can also point to a custom OpenAI-compatible endpoint.
Can I steer it to a topic?
Yes. The user-intent parameter says in words what to look for, and the model prioritizes those fragments when selecting.
Related
Skills and CLI from HeyGen: the agent writes HTML compositions and renders them into MP4 videos, captions, motion graphics and decks
Blender MCP
MCP for Blender
An MCP server and Blender addon: the agent creates and edits objects, materials and scenes, pulls assets and runs Python in Blender
Official Remotion skills for making videos in React: compositions, animation, captions, maps and rendering
Agentic video studio for Claude Code, Cursor and Codex: script, voiceover, images, stock footage, music, subtitles and render