Fetches YouTube video transcripts from a video URL or bare video ID. Provides tools to retrieve transcripts as plain text (paginated by character range) or as timestamped snippets with start times and durations (paginated by snippet index), supporting configurable caption language codes.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@mcp-youtube-transcript Summarize https://www.youtube.com/watch?v=dQw4w9WgXcQ"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
A small Model Context Protocol (MCP) server that fetches YouTube video transcripts so an AI assistant can read, summarize, or search what was said in a video.
It runs over stdio (for Claude Desktop, Claude Code, Cursor, and similar clients) or as a stateless Streamable HTTP server (for Docker or remote use).
Tool | What it does |
| Returns the transcript as one block of plain text. Paginates by character range. |
| Returns an array of |
Both tools accept a full YouTube URL (watch?v=, youtu.be/, /embed/, /v/) or a bare 11-character video ID.
get_transcriptArgument | Type | Default | Notes |
| string | required | YouTube URL or video ID |
| string |
| Caption language code |
| number |
| Character offset to start from |
| number |
| Max characters to return |
Response:
{
"text": "...",
"offset": 0,
"length": 12345,
"total_length": 12345
}get_timed_transcriptArgument | Type | Default | Notes |
| string | required | YouTube URL or video ID |
| string |
| Caption language code |
| number |
| Snippet index to start from |
| number |
| Max snippets to return |
Response:
{
"snippets": [{ "text": "hello", "start": 0.0, "duration": 1.2 }],
"offset": 0,
"count": 500,
"total_snippets": 1834
}Use total_length / total_snippets to decide whether to page for more.
Related MCP server: youtube-summarize
Requires Node.js 20 or newer.
git clone https://github.com/cchamp-msft/mcp-youtube-transcript.git
cd mcp-youtube-transcript
npm install
npm run buildThe server entry point is build/index.js.
claude mcp add youtube-transcript -- node /absolute/path/to/mcp-youtube-transcript/build/index.jsAdd to claude_desktop_config.json:
{
"mcpServers": {
"youtube-transcript": {
"command": "node",
"args": ["/absolute/path/to/mcp-youtube-transcript/build/index.js"]
}
}
}Other clients that support stdio MCP servers (Cursor, Windsurf, Zed, etc.) use the same command and args.
Then ask your assistant something like: "Summarize https://www.youtube.com/watch?v=dQw4w9WgXcQ".
Set MCP_TRANSPORT=http to expose a stateless Streamable HTTP endpoint instead of stdio.
MCP_TRANSPORT=http MCP_PORT=8000 npm startPath | Purpose |
| MCP endpoint |
| Returns |
Point an HTTP-capable client at http://localhost:8000/mcp.
docker build -t mcp-youtube-transcript .
docker run --rm -p 8000:8000 -e MCP_TRANSPORT=http mcp-youtube-transcriptAll configuration is via environment variables.
Variable | Default | Values |
|
|
|
|
| Port for HTTP mode |
|
|
|
npm run dev # tsc --watch
npm test # vitest
npm run build # compile to build/Source layout:
File | Role |
| Entry point, transport selection |
| MCP tool definitions |
| Video ID parsing and transcript fetching |
| Environment variable parsing |
Transcripts come from the unofficial youtube-transcript package, which scrapes YouTube's caption data. YouTube can change its page structure at any time and break it.
Videos without captions in the requested language return an error. Auto-generated captions usually work, but not always.
Heavy use from one IP address may be rate limited or blocked by YouTube.
HTTP mode is stateless and has no authentication. Do not expose it to the public internet without putting it behind a proxy that handles auth.
Issues and pull requests are welcome. For anything larger than a small fix, open an issue first so we can talk it over. Please run npm test and npm run build before submitting.
This server cannot be deployed
YouTube transcripts, search, channel browsing, and playlists for AI agents via MCP.
An MCP server that gives any LLM or agent clean YouTube transcripts on demand: a single video, a whole channel, or a playlist, plus AI cleanup of auto-generated captions. API-key auth, credit-based, same backend as the public v1 API. Get a free API key with 25 free credits at youtubetranscriptdownload.com/account.
SubDownload exposes YouTube as an MCP-native data source. Connect via OAuth and your AI agent can summarize videos, fetch full transcripts (even for videos with no captions, via AI ASR), search across channels, and save everything into a private knowledge base. Works with Claude, ChatGPT, Cursor, and 40+ MCP clients. Free credits on signup, no card required.
💯 The fastest YouTube transcript + YouTube search MCP for AI agents. Try for free.
