The Zoom Transcript MCP server acts as a smart bridge between Zoom meeting recordings and AI assistants. It allows users to bring their meeting conversations directly into their AI chat environment, making it possible to ask questions about past discussions, summarize key decisions, or find specific details from a video call without having to manually watch recordings or copy-paste text. By automating the retrieval of meeting data, it turns a library of Zoom recordings into a searchable, interactive knowledge base for any compatible LLM. Beyond simple fetching, this tool offers a robust set of features for managing transcript data. It can list available recordings within specific date ranges, identify participants, and automatically download transcripts from the most recent meetings. The server handles the heavy lifting of organizing these files into a structured system, categorizing them by month and including detailed metadata like meeting IDs, topics, and durations. This organization ensures that transcripts are not just downloaded, but are categorized in a way that remains easy for both humans and machines to navigate. For developers and technical users, the server provides four primary tools—`list_meetings`, `download_transcript`, `get_recent_transcripts`, and `search_transcripts`—which an AI model can invoke autonomously to find information. It utilizes Zoom’s Server-to-Server OAuth for secure authentication and stores data in a clean directory structure with both VTT transcript files and JSON metadata. This structured approach is particularly valuable for building RAG (Retrieval-Augmented Generation) workflows, as it allows an AI to perform targeted searches across large volumes of meeting history to provide contextually accurate answers based on real-world conversations.
Category: AI Memory & Context
Tags: meetings, oauth, rag, transcripts, zoom
bash git clone https://github.com/yourusername/zoom_transcript_mcp.git cd zoom_transcript_mcp 2. Install dependencies: bash npm install 3. Build the project: bash npm run buildmcpServers configuration file: json { "mcpServers": { "zoom-transcripts": { "command": "node", "args": ["/path/to/zoom-transcripts-server/build/index.js"], "env": { "ZOOM_ACCOUNT_ID": "your_zoom_account_id", "ZOOM_CLIENT_ID": "your_zoom_client_id", "ZOOM_CLIENT_SECRET": "your_zoom_client_secret", "TRANSCRIPTS_DIR": "/path/to/transcripts/directory" } } } } Obtaining Zoom Credentials: 1. Sign in to the Zoom App Marketplace. 2. Click "Develop" > "Build App" and choose Server-to-Server OAuth. 3. Under Scopes, add: * cloud_recording:read:list_account_recordings:admin * cloud_recording:read:recording:admin * cloud_recording:read:list_user_recordings:admin 4. Activate the app and collect your Account ID, Client ID, and Client Secret.list_meetings: Lists available Zoom meetings with recordings. * Parameters: dateRange (object with from and to), participant (optional string). * download_transcript: Downloads a transcript for a specific meeting. * Parameters: meetingId (string: Meeting ID or UUID). * get_recent_transcripts: Downloads transcripts from recent meetings. * Parameters: count (number: default 5). * search_transcripts: Searches across all downloaded transcripts for specific content. * Parameters: query (string), dateRange (optional object with from and to)."Search my Zoom transcripts for any mention of 'project timeline'." Manual tool call example for Claude: xml <use_mcp_tool> <server_name>zoom-transcripts</server_name> <tool_name>search_transcripts</tool_name> <arguments> { "query": "project timeline" } </arguments> </use_mcp_tool> Listing meetings within a date range: "List all my Zoom meetings with recordings from January 1st, 2025 to March 31st, 2025."search_transcripts tool, users can query their entire library of past Zoom discussions for specific keywords, technical terms, or project names. Example: "Search through all my meetings from the last three months for any mention of 'database migration' and tell me what the final decision was regarding the cut-over date."get_recent_transcripts tool can fetch the last several transcripts in one go. An AI can then synthesize these into a "Project Status Briefing," highlighting the evolution of a project's goals and current roadblocks. Example: "Download the transcripts for the last five 'Marketing Strategy' meetings and give me a high-level overview of how our social media strategy has changed over the past month."Part of MCP Servers
Zoom Transcript provides tools to list Zoom meetings with cloud recordings, download specific transcripts by meeting ID, pull recent transcripts automatically, and search the text of stored meetings. It also organizes VTT transcript files alongside JSON metadata by year and month.
Clone the GitHub repository, install dependencies using npm install, and run npm run build. Then register the built index.js script as an MCP server in your client configuration file, providing your Zoom Server-to-Server OAuth credentials via environment variables.
Zoom Transcript works with any client that implements the Model Context Protocol, including Claude Desktop, Cursor, and custom MCP host applications that support standard command-line server processes.
You need a Zoom account with cloud recording enabled and a Server-to-Server OAuth app configured with three scopes: cloud_recording:read:list_account_recordings:admin, cloud_recording:read:recording:admin, and cloud_recording:read:list_user_recordings:admin.
Yes, Zoom Transcript is open source and licensed under the MIT license. You can inspect, modify, and run the TypeScript source code from the official GitHub repository.