A curated collection of AI agent skills, each designed as a self-contained unit with a skill definition, supporting scripts, and templates.
Extract subtitles or transcribe Bilibili/YouTube videos, then generate polished Chinese lecture notes as a PDF handout.
Workflow:
- Fetch subtitles via
yt-dlpor transcribe withmlx-whisper - Read the transcript and metadata
- Draft comprehensive lecture notes in Chinese
- Render to a standalone PDF using LaTeX
Key features:
- Automatic subtitle reuse (skip re-downloading for previously processed videos)
- Bilingual cookie support for gated content
- Whisper fallback when subtitles are unavailable
- Subagent parallelism for long transcripts
- Clean, printable PDF output with styled boxes, code listings, and math support
Translate an arXiv paper or LaTeX source into a Chinese-only PDF while preserving its original structure and layout.
Workflow:
- Acquire and identify the paper's LaTeX source
- Replace visible English prose with polished Chinese
- Compile with XeLaTeX, repair layout issues, and clean temporary files
Key features:
- Preserves equations, figures, citations, labels, and document order
- Supports parallel translation of independent source files
- Delivers the final PDF with reproducible translated sources
AgentSkills/
├── README.md
├── README_zh.md # 中文说明
├── video-transcript-explainer/
│ ├── SKILL.md # Skill definition for Claude Code
│ ├── agents/ # Agent configurations
│ ├── scripts/ # Helper scripts
│ │ ├── video_transcript.py
│ │ └── render_notes_pdf.py
│ └── assets/ # Templates
│ └── notes-template.tex
└── translate-paper-zh/
│ ├── SKILL.md # Skill definition
│ └── agents/
│ └── openai.yaml # Agent interface metadata
- Python 3.10+
- uv
yt-dlpandmlx-whisper(installed viauv run --with ...)- Network access for retrieving arXiv LaTeX sources
latexmk+xelatexwith Chinese font support for PDF rendering