yt-transcript
drdoomsville/yt-transcript
Extraia e processe transcrições do YouTube sem API key
Ver no GitHubSkill para Claude Code que extrai transcrições de vídeos do YouTube diretamente de uma URL e salva em formatos Markdown e TXT. Oferece ações adicionais como resumo, pontos-chave, crítica e geração de perguntas de acompanhamento — sem necessidade de API key.
Instalação
npx convez-ai drdoomsville/yt-transcript1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Extrair transcrições completas de vídeos para análise posterior
- Gerar resumos e pontos-chave de conteúdo educacional ou técnico
- Criar críticas estruturadas sobre afirmações em vídeos
- Gerar perguntas de acompanhamento para aprofundamento em tópicos
Veredito

Ferramenta prática e bem-construída para automatizar a extração e processamento de conteúdo do YouTube. A arquitetura que delega o processamento para o agent é eficiente e elimina dependências de chaves de API, tornando a integração simples e confiável.
README
A tradução para o português deste README ainda está na fila.
yt-transcript
A Claude Code skill that extracts a YouTube
video's transcript from a URL and saves it as .md and .txt files — and
optionally applies an action (summarize, bullets, critique, follow-up
questions) to it.
The script only fetches and saves the transcript. When you ask for an action,
it hands the work to the agent (Claude) via an ACTION_REQUEST line — no LLM
runs in the script and no API key is needed, and the action wordings mirror
the yt-transcriber app.
Install
Clone into your Claude Code skills directory:
git clone git@github.com:drdoomsville/yt-transcript.git ~/.claude/skills/yt-transcript
Requirements:
- Python 3 on
PATH(or set$PYTHON/$env:PYTHON) pip install youtube-transcript-api- Network access (transcript API + YouTube's public oEmbed endpoint for the title)
Usage
Ask Claude naturally — e.g. "extract the transcript of <url>" or
"summarize this video: <url>" — and the skill runs. To invoke the script
directly:
# Just the transcript -> ./output/<title>.md and .txt
python ~/.claude/skills/yt-transcript/scripts/fetch.py "https://www.youtube.com/watch?v=VIDEO_ID"
# Transcript + an action
python ~/.claude/skills/yt-transcript/scripts/fetch.py --action summarize "https://www.youtube.com/watch?v=VIDEO_ID"
On Windows PowerShell, expand the home path manually:
python "$HOME\.claude\skills\yt-transcript\scripts\fetch.py" --action summarize "https://www.youtube.com/watch?v=VIDEO_ID"
Output lands under ./output/ relative to the current working directory.
Actions
--action <id> (alias -a). Valid ids:
| id | alias | produces |
|---|---|---|
summarize | summary | 3–5 paragraph summary, one-sentence TL;DR first |
bullets | 8–15 key points as a bullet list | |
criticize | critique | per-claim critique (supported / weak / counter-evidence) |
questions | 8–12 grouped follow-up questions |
With an action the script saves the transcript, then prints a final line:
ACTION_REQUEST {"action":"summarize","instruction":"…","transcript_file":"<abs .md>","output_file":"<abs .md>"}
The agent reads transcript_file, applies instruction, and writes the result
to output_file (e.g. output/<title>.summarize.md).
URL formats accepted
watch?v=<id>, youtu.be/<id>, /embed/<id>, /shorts/<id>, and raw 11-char
video IDs.
Notes
- Unknown action fails fast with a JSON error (before any network call).
- No captions / video unavailable exits 1 with a JSON error on stderr.
- Filenames strip characters illegal on Windows and truncate to 150 chars; on a title-fetch failure the 11-char video ID is used instead.
- Existing files at the same path are overwritten without prompt.
See SKILL.md for the full agent-facing instructions.
License
MIT © 2026 Sonny Kocak