Tubescribe

MCP server for fetching YouTube video transcripts. Thin wrapper around youtube-transcript-api.

No API key required.

Install

pip install tubescribe-mcp

Usage

tubescribe-mcp

The server runs on stdio transport (MCP standard).

Tools

get_transcript(video, languages?, timestamps?) — fetch a video transcript as plain text.

• video — YouTube URL or bare 11-char video ID
• languages — comma-separated preference codes, e.g. "en,de" (default: English)
• timestamps — prefix each line with [MM:SS] (default: false)

list_transcript_languages(video) — list available transcript languages for a video.

Library

from tubescribe_mcp.transcript import fetch_transcript, list_languages, extract_video_id

text = fetch_transcript("https://www.youtube.com/watch?v=dQw4w9WgXcQ")
langs = list_languages("dQw4w9WgXcQ")
vid = extract_video_id("https://youtu.be/dQw4w9WgXcQ")

License

GPL-3.0-or-later