用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/tomevault-io/skills-registry --skill youtube命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
| Use when this capability is needed.
> Use when this capability is needed.
Review architecture and API design for the vfs-s3 project. Use when the user mentions @architect, asks to review an issue's design, discuss module boundaries, API shape, or architectural decisions for vfs-s3. Also trigger when the user wants to create an ADR (Architecture Decision Record) or evaluate a technical approach for the project. Intended for dispatch from Codex automation or Claude routines; GitHub trigger phrase: @vfs-s3-bot please prepare design doc Use when this capability is needed.
基于 SOC 职业分类
正在显示 SKILL.md
| name | youtube |
| description | | Use when this capability is needed. |
The most comprehensive YouTube skill for AI agents.
We analyzed 15+ YouTube MCP servers and found each does one thing well, but none does everything. So we built the skill we wished existed.
| What Others Do | What We Do |
|---|---|
| Transcripts OR search OR downloads | All three, unified |
| Burn API quota on transcripts | FREE transcripts (zero quota) |
| Single video at a time | Batch operations (50 videos) |
| Basic search | Filtered search (date, duration, order) |
| Text output only | JSON export for pipelines |
Most tools use the YouTube Data API for transcripts = 100 quota units per request. Daily limit is 10,000 units = only ~100 transcripts/day.
We use youtube-transcript-api — extracts directly from YouTube's frontend. Zero API quota. Unlimited transcripts.
| Command | Quota | What it does |
|---|---|---|
transcript VIDEO | FREE | Get video transcript |
transcript-list VIDEO | FREE | List available languages |
download VIDEO | FREE | Download video (yt-dlp) |
download-audio VIDEO | FREE | Extract audio only |
search QUERY | 100 | Search videos |
video ID [ID...] | 1/video | Get details (batch!) |
comments VIDEO | 1 | Get comments + replies |
channel [ID] | 1-3 | Channel statistics |
# 1. Get credentials from Google Cloud Console
# - Create OAuth 2.0 Client ID (Desktop app)
# - Download JSON
# 2. Save credentials
mkdir -p ~/.config/youtube-skill
mv ~/Downloads/client_secret*.json ~/.config/youtube-skill/credentials.json
# 3. Authenticate
uv run {baseDir}/scripts/youtube.py auth
# Plain text transcript
uv run {baseDir}/scripts/youtube.py transcript VIDEO_ID
# With timestamps
uv run {baseDir}/scripts/youtube.py transcript VIDEO_ID --timestamps
# Specific language (falls back to available)
uv run {baseDir}/scripts/youtube.py transcript VIDEO_ID -l es
# List what's available
uv run {baseDir}/scripts/youtube.py transcript-list VIDEO_ID
# JSON output
uv run {baseDir}/scripts/youtube.py transcript VIDEO_ID --json
Works with URLs too:
uv run {baseDir}/scripts/youtube.py transcript "https://youtube.com/watch?v=dQw4w9WgXcQ"
# Basic search
uv run {baseDir}/scripts/youtube.py search "AI news 2026"
# With filters
uv run {baseDir}/scripts/youtube.py search "tutorial" -l 20 --order date
uv run {baseDir}/scripts/youtube.py search "lecture" --duration long
uv run {baseDir}/scripts/youtube.py search "news" --published-after 2026-01-01T00:00:00Z
# Single video
uv run {baseDir}/scripts/youtube.py video dQw4w9WgXcQ
# Multiple videos at once (up to 50)
uv run {baseDir}/scripts/youtube.py video id1 id2 id3 id4 id5
# JSON output for processing
uv run {baseDir}/scripts/youtube.py video id1 id2 --json
# Top comments
uv run {baseDir}/scripts/youtube.py comments VIDEO_ID
# With replies
uv run {baseDir}/scripts/youtube.py comments VIDEO_ID --replies
# Recent comments
uv run {baseDir}/scripts/youtube.py comments VIDEO_ID --order time -l 50
# Video (best quality)
uv run {baseDir}/scripts/youtube.py download VIDEO_ID
# Specific resolution
uv run {baseDir}/scripts/youtube.py download VIDEO_ID -r 720p
# With subtitles
uv run {baseDir}/scripts/youtube.py download VIDEO_ID -s en
# Audio only (MP3)
uv run {baseDir}/scripts/youtube.py download-audio VIDEO_ID
# Audio as M4A
uv run {baseDir}/scripts/youtube.py download-audio VIDEO_ID -f m4a
uv run {baseDir}/scripts/youtube.py subscriptions
uv run {baseDir}/scripts/youtube.py playlists
uv run {baseDir}/scripts/youtube.py playlist-items PLAYLIST_ID
uv run {baseDir}/scripts/youtube.py liked
uv run {baseDir}/scripts/youtube.py channel
| Full | Alias |
|---|---|
transcript | tr |
search | s |
video | v |
comments | c |
download | dl |
download-audio | dla |
Research: Fetch transcript → analyze with LLM → extract insights
Learning: Batch transcripts from playlist → create study notes
Monitoring: Search recent videos → extract transcripts → track trends
Podcasts: Download audio for offline listening
Analysis: Get channel stats → compare competitors
uv run {baseDir}/scripts/youtube.py -a work subscriptions
uv run {baseDir}/scripts/youtube.py -a personal liked
We surveyed the landscape:
None combined free transcripts + search + downloads + batch ops.
Now one does.
Converted and distributed by TomeVault — claim your Tome and manage your conversions.