# YouTube Transcript AI [Health: Active]

**Category:** 💻 Developer Tools  
**Repository:** https://github.com/zxl777/youtube-transcript-mcp  
**GitHub Stars:** 0  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/youtube-transcript-ai

## Description
Fetch the full transcript of any YouTube video as clean text. No API key, no signup.

## Tools
Capabilities this server exposes over MCP:

- **get_youtube_transcript** — Fetch the full text transcript of a YouTube video.

## Claude Desktop Quick Installation
Remote MCP endpoint (confidence: high). Install path detected from listing signals. Add as a URL/SSE server in your client:

```json
"mcpServers": {
  "youtube-transcript-ai": {
    "url": "https://youtube-transcript.ai/mcp"
  }
}
```

## Documentation

## What YouTube Transcript AI does

The YouTube Transcript AI MCP server gives compatible AI clients access to captions from YouTube videos. It exposes one MCP tool, `get_youtube_transcript`, which returns the complete transcript as plain text rather than requiring an agent to open a browser page or manually copy captions.

The service is hosted at `https://youtube-transcript.ai/mcp`. It is intended for MCP clients that can connect to a remote Streamable HTTP endpoint, as well as clients that can use `mcp-remote` as a bridge. The README identifies Claude, ChatGPT, Cursor, Windsurf, Cline, and Claude Desktop as supported setup targets; ChatGPT and Claude can use the remote connector configuration directly, while several desktop and editor clients can use the supplied `mcp-remote` configuration.

## How it works

The YouTube Transcript AI MCP server receives a YouTube link or a bare 11-character video ID through the tool call. Accepted URL forms include watch URLs, `youtu.be` links, Shorts URLs, and embed URLs. An optional `lang` value selects a caption track with a BCP-47 language code such as `en`, `zh-CN`, `es`, or `ja`. If that parameter is omitted, the service uses the default track.

The response is plain text containing the transcript, along with the video title, duration, and a list of available languages. The hosted service fetches captions from publicly available sources. Its README also describes proxy fallback in the fetch pipeline for videos that block typical scrapers, but availability still depends on the source video and the service endpoint.

## Setup and configuration

No local installation or API key is required for the hosted setup. Configure the MCP endpoint as:

`https://youtube-transcript.ai/mcp`

Clients with native remote MCP support can use that URL directly. For clients that launch MCP processes from a configuration file, the documented approach runs `npx` with `mcp-remote` and passes the same endpoint URL. A representative configuration uses the server name `youtube-transcript` and does not define environment variables or credentials.

The service also speaks standard MCP over Streamable HTTP using JSON-RPC. A Node.js quickstart is provided in the repository for programmatic use, and the README shows it being invoked with a YouTube URL.

## Tools and capabilities

`get_youtube_transcript` is the only exposed tool. It has two inputs:

- `video` — required string containing a supported YouTube URL or an 11-character video ID.
- `lang` — optional string containing the desired BCP-47 caption language code.

The YouTube Transcript AI MCP server can support agent workflows such as summarizing a video, translating its transcript, locating discussion of a topic, or extracting exact lines for a response. These tasks are performed by the connected AI client after the transcript is returned; the MCP tool itself retrieves the transcript and metadata.

## Limitations and notes

The endpoint is rate-limited by IP to help keep the hosted service free. The README recommends contacting the service for heavy or automated workloads. It does not provide an API-key or account-based configuration path.

Transcript access is based on publicly available captions, so a requested video must have captions that the service can retrieve. Language selection only works when a corresponding caption track is available. The service is not affiliated with YouTube or Google, and the repository is licensed under MIT.

_Full upstream README: https://allmcps.com/mcp/youtube-transcript-ai/readme_