The YouTube Insights MCP Server acts as a bridge between YouTube's massive library of video content and AI assistants. It allows users to quickly pull information from videos without having to watch them manually. Whether someone needs to read a video's transcript, find specific videos based on a topic, or look up details about a YouTube channel, this tool makes that data instantly accessible to their AI agent for analysis. For more technical workflows, the server provides three specialized tools that streamline data extraction through the Model Context Protocol. The `get_youtube_transcript` tool fetches full subtitles in multiple languages, such as English and Korean, making it easy for an LLM to summarize long videos or find specific quotes. The `search_youtube_videos` tool enables programmatic video discovery based on keywords, returning rich metadata like view counts, likes, and thumbnails. Additionally, the `get_channel_info` tool allows agents to retrieve a creator's profile data, including subscriber counts and recent uploads, to provide a broader context for the content being analyzed. Developers building AI-powered applications will find this MCP server easy to deploy and integrate. It is built using the FastMCP framework and can be installed automatically via Smithery or run directly using the `uvx` tool. By providing a YouTube Data API key, developers can empower their LLMs to "watch" and analyze video content in real-time. This capability is essential for building research agents, automated content curation tools, or competitive analysis systems that require reliable access to YouTube’s global data.
Category: Design, Media & Creative
Tags: metadata, search, transcripts, video, youtube
Visit YouTube Insights MCP Server
bash npx -y @smithery/cli install @dabidstudio/youtubeinsights-mcp-server --client claude Using uvx (Recommended) No specific installation is needed when using uvx. You can run it directly via configuration. Development Installation 1. Clone the GitHub repository. 2. Copy .env.example to .env and enter your YouTube Data API credentials (YOUTUBE_API_KEY). ---claude_desktop_config.json for Claude Desktop). Standard Configuration (Using uvx): json { "mcpServers": { "youtubeinsights": { "command": "uvx", "args": ["youtubeinsights-mcp-server"], "env": { "YOUTUBE_API_KEY": "your-api-key" } } } } Development Configuration: json { "mcpServers": { "youtubeinsights": { "command": "uv", "args": [ "--directory", "path/to/youtubeinsights-mcp-server", "run", "youtubeinsights-mcp-server" ], "env": { "YOUTUBE_API_KEY": "your-api-key" } } } } ---get_youtube_transcript: Extract full transcript (subtitles) from a YouTube video URL. Supports multiple languages (e.g., ko, en). * search_youtube_videos: Search for videos on YouTube by keyword and retrieve metadata including views, likes, and thumbnails. * get_channel_info: Retrieve channel metadata (title, subscriber count, etc.) and latest uploads based on any YouTube video URL. ---get_youtube_transcript tool, an AI agent can pull the full text of a video. The AI can then process this transcript to create structured notes, highlight key takeaways, or answer specific questions about the content without the user ever pressing "play." Example: A developer provides a 40-minute Kubernetes tutorial URL. The AI uses the MCP to fetch the transcript, identifies the specific CLI commands mentioned, and generates a "Cheat Sheet" summary for the developer to use.search_youtube_videos and get_channel_info. Users can search for specific keywords to see which videos are currently "winning" the algorithm based on metadata like view counts and engagement. Example: A marketing manager asks, "What are the top 5 most-viewed videos about 'Generative AI' from the last month, and how many subscribers do those channels have?" The AI uses the search tool to find the videos and the channel tool to gather subscriber data, presenting a comparative report.get_channel_info tool, an agent can take a single video URL and instantly retrieve the channel’s subscriber count and a list of their latest uploads. This helps in assessing the creator’s consistency and current audience size. Example: A PR agent finds a viral tech review. They ask the AI, "Check this video's creator—how many subscribers do they have, and what else have they posted recently?" The MCP fetches the channel profile and recent video list to help the agent decide whether to send a collaboration pitch.Part of MCP Servers