Back to search
skills
SkillWorks with Claude CodeGitHub
Claude Code skills — live-stream-clipper: auto-clip Twitch/YouTube streams on chat hype into captioned vertical Shorts
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Badge for your README
Markdown
[](/listings/morphhyy--skills)HTML
<a href="/listings/morphhyy--skills"><img src="/api/listings/morphhyy--skills/badge.svg" alt="findsafeskills safety rating" /></a>Install
i
Install instructions from the README
As written by the repo's author. findsafeskills has not verified these steps.
git clone https://github.com/morphhyy/skills.git cp -r skills/live-stream-clipper ~/.claude/skills/ # every project # or cp -r skills/live-stream-clipper <your-project>/.claude/skills/ # one project
Restart Claude Code. The skill loads on its own when you ask for something it covers, for example *"clip xQc's stream live and give me 3 shorts"*.
---
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
README.md · README · 139 lines
- # skills
- Skills for [Claude Code](https://claude.com/claude-code). Each folder is one skill: a `SKILL.md` that tells Claude when and how to use it, plus any scripts it runs.
- | Skill | What it does |
- |---|---|
- | [live-stream-clipper](live-stream-clipper/SKILL.md) | Clips Twitch and YouTube streams **while they're live**. Watches chat for hype spikes, cuts the moment, burns in captions, renders a 1080×1920 Short, and can upload it to YouTube. |
- ## Install a skill
- ```bash
- git clone https://github.com/morphhyy/skills.git
- cp -r skills/live-stream-clipper ~/.claude/skills/ # every project
- # or
- cp -r skills/live-stream-clipper <your-project>/.claude/skills/ # one project
- ```
- Restart Claude Code. The skill loads on its own when you ask for something it covers, for example *"clip xQc's stream live and give me 3 shorts"*.
- ---
- ## live-stream-clipper
- ```
- streamlink / yt-dlp ──► ffmpeg 10s segment buffer
- Twitch IRC / YouTube live chat ──► spike detector ──► cut [spike-22s, spike+8s]
- ──► whisper.cpp captions ──► vertical Short ──► (optional) YouTube upload
- ```
- Everything is free. Chat is read anonymously. Only the YouTube upload needs a Google account.
- ### Requirements
- | Tool | Used for |
- |---|---|
- | Python 3.10+ | the clipper (standard library only) |
- | [ffmpeg](https://ffmpeg.org) with libass and freetype | recording, cutting, captions, covers |
- | [streamlink](https://streamlink.github.io) | Twitch recording |
- | [yt-dlp](https://github.com/yt-dlp/yt-dlp) | YouTube recording and live chat |
- | [whisper.cpp](https://github.com/ggml-org/whisper.cpp) (`whisper-cli`) + a model | captions (optional: use `--no-captions`) |
- | [uv](https://docs.astral.sh/uv/) | YouTube uploader (pulls its own Google API packages) |
- **macOS**
- ```bash
- brew install python ffmpeg streamlink yt-dlp whisper-cpp uv
- ```
- **Debian / Ubuntu**
- ```bash
- sudo apt install python3 ffmpeg
- curl -LsSf https://astral.sh/uv/install.sh | sh
- uv tool install streamlink && uv tool install yt-dlp
- # whisper-cli: build from source (https://github.com/ggml-org/whisper.cpp#quick-start)
- ```
- **Whisper model** (about 466 MB; use `base` for a smaller and faster model)
- ```bash
- mkdir -p ~/.cache/whisper
- curl -L -o ~/.cache/whisper/ggml-small.bin \
- https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-small.bin
- ```
- > Tested on macOS. Linux should work. Windows isn't supported yet: the recorder's process handling and the helper `.sh` scripts assume a POSIX system, so use WSL.
- ### Usage
- Run these from your project folder. Clips are saved there.
- ```bash
- S=~/.claude/skills/live-stream-clipper/scripts
- python3 $S/clipper.py xqc # Twitch channel
- python3 $S/clipper.py "https://www.youtube.com/watch?v=VIDEO_ID" # YouTube live
- python3 $S/clipper.py xqc --max-clips 3 --export output # stop after 3, copy to ./output
- python3 $S/clipper.py xqc --layout split --cam 0:0:480:270 # facecam on top, gameplay below
- python3 $S/clipper.py xqc --upload --privacy private # upload every Short
- ```
- It waits for the stream to go live, learns the normal chat speed for 90 seconds, then clips when chat spikes. It survives stream drops and keeps only the last 30 minutes of recording on disk.
- - **Watch progress:** `tail -f clips-xqc/events.log`
- - **Force a clip now:** `touch clips-xqc/CLIP_NOW`
- - **Stop:** Ctrl+C
- Output:
- ```
- clips-<channel>/
- events.log LIVE / SPIKE / CLIP READY / UPLOAD
- chat.jsonl every chat message with timestamps
- clips/<id>/ raw.mp4 short.mp4 captions.srt meta.json
- ```
- **Layouts:** `blur` (default; the full frame over a blurred background), `crop` (centre crop, good for IRL), `split` (facecam above gameplay; give the facecam box with `--cam x:y:w:h` in source pixels).
- ### Tuning
- | Problem | Try |
- |---|---|
- | Too many clips | `--mult 4 --floor 40 --cooldown 90` |
- | No clips (small chat) | `--mult 2.5 --floor 6` |
- | Clip starts too late | raise `--pre` (default 22s) |
- | Clip ends too early | raise `--post` (default 8s) |
- | Wrong caption language | `--lang en` |
- | Captions too slow | use the `base` model with `--model ~/.cache/whisper/ggml-base.bin` |
- Run `python3 $S/clipper.py --help` for every option.
- ### YouTube upload (optional, one time)
- 1. In [Google Cloud Console](https://console.cloud.google.com), create a project and enable **YouTube Data API v3**.
- 2. Set up the **OAuth consent screen** as External and add yourself as a test user.
- 3. Create an **OAuth client ID** of type **Desktop app** and save the JSON as `~/.config/live-clipper/client_secret.json`.
- 4. Log in once: `uv run $S/upload_short.py --auth`
- Upload a single clip by hand:
- ```bash
- uv run $S/upload_short.py --meta clips-xqc/clips/<id>/meta.json --title "your title" --privacy private
- ```
- **Limits:**
- - **Private only at first.** Google forces uploads from unaudited API projects to private. Make them public in YouTube Studio, or apply for the API audit.
- - **About 6 uploads a day.** Each upload costs about 1,600 of the default 10,000 daily quota units.
- ### Credentials stay local
- The OAuth client, login token and any cookies live in `~/.config/live-clipper/`, never in this repo. The token is written with `0600` permissions. If YouTube blocks the recorder as a bot, export cookies to `~/.config/live-clipper/yt-cookies.txt`. Each run copies them into its own folder, so don't commit or share your `clips-*` folders.
- ### Use responsibly
- These clips come from someone else's stream. Credit the streamer (the default description does), and get their permission or join their clipping program before you upload. Copyright claims and strikes land on **your** channel, and game footage can trigger Content ID.
- ## License
- [MIT](LICENSE)
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.