get_youtube_transcript

Pulls a transcript and full video metadata.

Input schema

{
  video: string  // YouTube URL or 11-character video ID
}

Field	Type	Description
`video`	string	URL (`watch?v=`, `youtu.be/`, `embed`, `shorts`, `live`) or bare ID.

Output

{
  videoId: string
  title: string
  channel: string
  durationSeconds: number
  language: string | null
  transcript: string         // joined newline-separated text
  segments: Array<{
    start: number            // seconds
    end: number              // seconds
    text: string
  }>
  source: 'innertube' | 'ytdlp'
}

Example agent prompt

Pull the transcript for https://youtube.com/watch?v=dQw4w9WgXcQ and tell me the three biggest lyrics themes.

The agent picks get_youtube_transcript, the MCP client makes the OAuth-authenticated call, and the agent reasons over the returned text.

Errors

Errors come back as isError: true results with a text content payload like:

transcript_unavailable: Video exists but has no transcript.

See Errors for the full code list.

Plan

Free plan, 1 quota per call (cached calls don't count).