agentsclimarketplace
Sign in

Youtube transcript

Skill nord342/claude-youtube-transcript/plugins/youtube-transcript/skills/youtube-transcript

Fetch the transcript/captions of any YouTube video. Use whenever the user pastes a YouTube link (youtube.com, youtu.be, shorts) or asks for a video's transcript, subtitles, captions, or "what does this video say / summarize this video". Runs locally via yt-dlp with browser cookies, so it works reliably without an API key.From its SKILL.md

Install
npx -y skills add nord342/claude-youtube-transcript --skill youtube-transcript

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

3.4 KB, 767 tokens by cl100k_base, as published. Nobody here has run it

YouTube Transcript

Get the full transcript of a YouTube video so you can summarize it, quote it, answer questions about it, or pull out key points — all locally, no API key.

When to use

Trigger this skill whenever the user:

  • Pastes any YouTube URL (youtube.com/watch?v=, youtu.be/, /shorts/, /embed/), with or without a question.
  • Asks for a video's transcript, captions, or subtitles.
  • Asks you to summarize, analyze, quote, or answer questions about a YouTube video.

How to fetch a transcript

Run the bundled script. It calls yt-dlp with your browser cookies (which is what reliably gets past YouTube's bot checks), reads the caption track, and prints clean text to stdout. Progress/metadata go to stderr.

python3 "${CLAUDE_PLUGIN_ROOT}/skills/youtube-transcript/scripts/yt_transcript.py" "<YOUTUBE_URL>"

If ${CLAUDE_PLUGIN_ROOT} is not set in your environment, use the script's path inside this skill directory (scripts/yt_transcript.py).

Options

  • --timestamps — prefix each line with [mm:ss] (use when the user wants to cite or jump to moments).
  • --lang <code> — preferred caption language, e.g. --lang es (default en; falls back to English, then any available).
  • --list-langs — list which caption languages exist for the video, then exit. Use this if the default language isn't what the user wanted.
  • --browser <name> — cookie source: auto (default), chrome, safari, brave, edge, firefox. auto tries each, then no cookies.
  • --json — emit {title, uploader, duration_seconds, language, caption_source, transcript} as JSON. Use when you want the metadata too.

Typical flow

  1. Run the script with the user's URL. For long videos, prefer --json or plain text and then work from the captured output.
  2. Lead your answer with the video title and channel (shown on stderr, or in --json), so the user knows it fetched the right video.
  3. Then do what the user asked — summarize, extract steps/quotes, answer a question, etc. If they didn't ask for anything specific, give a concise summary and offer the full transcript.
  4. Don't paste a huge raw transcript back unless the user explicitly asks for the full text.

Troubleshooting

  • "yt-dlp is not installed" → brew install yt-dlp (or pipx install yt-dlp).
  • Metadata fetch fails / "Sign in to confirm you're not a bot" → the user must be signed in to YouTube in the chosen browser. Try --browser chrome explicitly, or yt-dlp -U to update. A logged-in Chrome profile is the most reliable on macOS.
  • "no captions/subtitles available" → the video genuinely has no captions (some do not). There is no transcript to fetch; tell the user.
  • TLS / certificate errors → the script auto-falls back, but the clean fix is pip3 install certifi.

Notes

  • No video is downloaded — only the caption track. It's fast.
  • Prefers human-made (manual) subtitles over auto-generated captions when both exist.
  • Works on unlisted videos you can access; will not work on private/members-only videos your browser can't see.

What ships with it: 1 file

10.7 KB alongside SKILL.md, 1 of them executable

scripts/

Gives 0 of the 12 instructions most video audio skills give in 767 tokens

Counted across 622 of the 795 authors here whose files we hold, read 2026-08-07

  • Read individual rule files for detailed explanationsin 21 of 622, across 10 files
  • Render final videoin 13 of 622, across 6 files
  • Use WAV PCM 16kHz mono audio formatin 12 of 622, across 3 files
  • Use this skill when dealing with Remotion codein 11 of 622, across 4 files
  • Save generated audio to a WAV filein 11 of 622, across 4 files
  • Handle conversion errors gracefullyin 10 of 622, across 6 files
  • Add captions to videos alwaysin 10 of 622, across 4 files
  • Generate music from text descriptions using MusicGenin 9 of 622, across 2 files
  • Do not skip pipeline layersin 9 of 622, across 3 files
  • Do not make one tool do everythingin 9 of 622, across 3 files
  • Use Azure Document Intelligence for complex PDFsin 9 of 622, across 4 files
  • Never ask the user to paste their full API keyin 9 of 622, across 3 files

Said here and by no other author read

  • trigger on any provided YouTube URL
  • lead the answer with the video title and channel
  • fulfill the user's specific request after fetching
  • provide a concise summary if no specific request is made

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 326,499. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.