Works from anywhere. Runs fine on AWS, Lambda, Vercel, n8n Cloud, and other cloud hosts. There are no proxies to buy and no IP bans to work around. YouTube blocks most cloud IPs for transcript requests; the yTranscript API handles that for you.
Handles videos without captions. If a video has no captions, it's transcribed with AI speech-to-text.
Returns the language actually spoken by default, including on auto-dubbed videos. Pass lang to pick a specific caption track.
Keeps long videos manageable. Choose plain text or timestamped lines, and cap the length so a two-hour video doesn't flood the model's context.

Setup

Get a free API key at ytranscript.com/developers. The free plan includes 50 units a month, no card required.
Add the server to your client using one of the configs below.

Claude Code

claude mcp add ytranscript -e YTRANSCRIPT_API_KEY=yk_live_your_key -- npx -y ytranscript-mcp

Claude Desktop

Add this to claude_desktop_config.json. Open it from Settings → Developer → Edit Config.

{
  "mcpServers": {
    "ytranscript": {
      "command": "npx",
      "args": ["-y", "ytranscript-mcp"],
      "env": { "YTRANSCRIPT_API_KEY": "yk_live_your_key" }
    }
  }
}

Cursor

Add this to ~/.cursor/mcp.json, or to .cursor/mcp.json in a project:

{
  "mcpServers": {
    "ytranscript": {
      "command": "npx",
      "args": ["-y", "ytranscript-mcp"],
      "env": { "YTRANSCRIPT_API_KEY": "yk_live_your_key" }
    }
  }
}

VS Code

Add this to .vscode/mcp.json:

{
  "servers": {
    "ytranscript": {
      "command": "npx",
      "args": ["-y", "ytranscript-mcp"],
      "env": { "YTRANSCRIPT_API_KEY": "yk_live_your_key" }
    }
  }
}

Any other MCP client that runs stdio servers works the same way: run npx -y ytranscript-mcp with YTRANSCRIPT_API_KEY set.

Tool: `get_transcript`

Argument	Required	Description
`video`	yes	A YouTube URL (`watch`, `youtu.be`, `shorts`, `embed`) or an 11-character video ID
`lang`	no	Language code of the caption track to fetch, such as `en` or `es`. Omit it to get the video's original language
`format`	no	`text` (default) returns plain text. `timestamped` returns one line per segment with `[m:ss]` times
`max_characters`	no	Cut the transcript at about this many characters. The response says how much was left out