This YouTube Transcript MCP acts as a bridge between AI models and video content, allowing users to pull text transcripts directly from any YouTube video. It simplifies the process of summarizing lectures, extracting key points from tutorials, or searching through video dialogue by providing a clean text version of the audio. Because it is a "remote" server, it works across all platforms—including mobile devices—without requiring users to install complex software or scripts on their own machines. Beyond simple extraction, the tool is designed to be highly versatile and robust. It automatically recognizes any YouTube link format, including standard URLs, shortened links, mobile addresses, and even YouTube Shorts or Live recordings. It also supports multi-language transcript retrieval, enabling users to specify a preferred language code for international content. To ensure speed and reliability, the server employs a smart caching system that saves previously requested transcripts for seven days, drastically reducing wait times for popular content. For developers and LLM integrators, this server is unique in offering a remote-first Model Context Protocol experience using Server-Sent Events (SSE) or HTTP transport. It is hosted on Cloudflare Workers, ensuring high availability and global low-latency performance through its edge network. Developers can easily integrate the `get_transcript` tool into their AI configurations, allowing an LLM to programmatically fetch data with built-in error handling and exponential backoff for rate limiting. This zero-maintenance architecture removes the friction of managing local dependencies, making it an ideal plug-and-play component for building sophisticated AI agents that need to process video information.
Category: Design, Media & Creative
Tags: cloudflare, summarization, transcription, video, youtube
npx directly in your configuration. Option 2: Self-Hosted Deployment If you wish to run your own instance on Cloudflare Workers, use the following commands: bash git clone https://github.com/ergut/youtube-transcript-mcp cd youtube-transcript-mcp npm install npm run deploy ---claude_desktop_config.json file: Standard Configuration: json { "mcpServers": { "youtube-transcript": { "command": "npx", "args": [ "mcp-remote", "https://youtube-transcript-mcp.ergut.workers.dev/sse" ] } } } Advanced Configuration (with Debug Logging): json { "mcpServers": { "youtube-transcript": { "command": "npx", "args": [ "mcp-remote", "https://youtube-transcript-mcp.ergut.workers.dev/sse", "--debug" ] } } } Advanced Configuration (HTTP Transport Only): json { "mcpServers": { "youtube-transcript": { "command": "npx", "args": [ "mcp-remote", "https://youtube-transcript-mcp.ergut.workers.dev/mcp", "--transport", "http-only" ] } } } ---get_transcript * Description: Extracts the transcript from a YouTube video URL. * Parameters: * url (required): The YouTube video URL (supports all standard formats including shorts, live, and mobile). * language (optional): Language code for the transcript (e.g., 'en', 'tr', 'es'). Defaults to 'en'. ---language parameter allows users to specify which transcript version to pull. Once the transcript is in the chat context, Claude’s native translation and reasoning capabilities can be used to translate, summarize, and explain foreign-language content with high nuance. Example: A researcher finds a technical talk in Japanese regarding new battery technology. They use get_transcript(url="...", language="ja") and then tell Claude: "Translate this transcript into English and highlight any mentions of 'solid-state' energy density specs."Part of MCP Servers
No local installation is required for the public server. You can configure it directly in your Claude Desktop configuration file by invoking npx with mcp-remote pointed at the hosted endpoint. If you prefer to self-host, clone the GitHub repository, install the npm dependencies, and deploy the project to Cloudflare Workers using the npm run deploy script.
The server provides the get_transcript tool to extract spoken text captions from YouTube URLs. It processes standard links, shortened links, mobile links, and Shorts, automatically strips tracking parameters, and supports language selection via codes such as en, es, tr, or ja. Extracted transcripts are cached in Cloudflare KV for seven days.
It works with any MCP-compliant client that supports Server-Sent Events or HTTP transport via the mcp-remote bridge. This includes Claude Desktop across macOS and Windows, as well as mobile and web configurations using compatible MCP proxy setups or custom JSON-RPC endpoints.
Yes, YouTube Transcript is an open-source project. The source code is publicly hosted on GitHub and distributed under the MIT License, allowing developers to review the codebase, submit issues or pull requests, and deploy custom instances on their own Cloudflare Workers accounts.