Krisp MCP

Connect your AI tools to your Krisp workspace using the Model Context Protocol (MCP), an open standard that lets AI assistants interact with your meeting data, action items, and calendar.
 

Overview

Krisp MCP is a hosted server that gives AI tools secure access to your Krisp workspace. It is designed to work with AI assistants like Claude Code, Cursor, VS Code, ChatGPT, and others.

Krisp MCP supports Streamable HTTP transport only. Server-Sent Events (SSE) is not supported.
 

Why use Krisp MCP?

• Easy setup: Connect through OAuth 2.0 with PKCE; complete the flow when your tool prompts you
• Meeting intelligence: AI tools can search meetings, read transcripts and summaries, and list action items
• Optimized for AI: Tools return structured data and natural-language summaries for agent use

What can you do with Krisp MCP?

• Search meetings: Find past meetings by topic, content, attendees, or date range
• Read meeting content: Fetch full transcripts, summaries, key points, and action items by document ID
• Manage tasks: List action items, filter by assignee or completion status
• Plan ahead: List upcoming calendar meetings and agenda IDs for the next 1–14 days
• Stay updated: List Activity Center notifications (assignments, summaries, etc.)
• Date/time: Get current date/time or enumerate dates in a range for accurate scheduling

Connecting to Krisp MCP

This guide walks you through connecting your AI tool to Krisp MCP. Once connected, your tool can read meeting data from your Krisp workspace based on your access and permissions.
 

Transport and URL

Cursor

1. Open Cursor Settings → MCP → Add new global MCP server

2. Paste the following configuration (JSON):

   {
        "mcpServers": {
          "krisp": {
            "url": "https://mcp.krisp.ai/mcp"
          }
        }
      }

3. Save and restart Cursor. When you use a Krisp tool for the first time, complete the OAuth flow to connect your workspace

Project-level configuration

To share the Krisp MCP configuration with your team, create a ".cursor/mcp.json" file in your project root:

{
  "mcpServers": {
    "krisp": {
      "url": "https://mcp.krisp.ai/mcp"
    }
  }
}

Claude Code

Run the following in your terminal:

claude mcp add --transport http krisp https://mcp.krisp.ai/mcp

Then authenticate by running "/mcp" in Claude Code and following the OAuth flow.

Other tools

If your AI tool supports MCP with a remote HTTP server, use:

{
  "mcpServers": {
    "krisp": {
      "url": "https://mcp.krisp.ai/mcp"
    }
  }
}

If your tool does not support remote HTTP connections directly, use the mcp-remote bridge:

{
  "mcpServers": {
    "krisp": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://mcp.krisp.ai/mcp"]
    }
  }
}

Security best practices

• Use the official endpoint only: Connect only to https://mcp.krisp.ai/mcp. Do not use other URLs for Krisp MCP
• Review actions: Enable human confirmation when possible so you can review what the agent does with your data
• Protect your data: Untrusted tools or agents could be instructed to exfiltrate data. Use only trusted MCP clients and review permissions
• Secure token storage: Store access and refresh tokens securely (e.g. encrypted); never expose them in client-side or public storage
• HTTPS only: Use HTTPS for all OAuth redirect URIs and MCP connections in production

Troubleshooting

• My tool doesn't support remote MCP servers: Use the mcp-remote configuration above.
• Authentication issues: Complete the OAuth flow when prompted; disconnect and reconnect if access is lost; ensure you have the right permissions in your Krisp workspace.
• Bearer token is missing: Requests must include `Authorization: Bearer <access_token>`. Complete the OAuth flow and ensure your client sends the token on each request.
• Document not found: Document IDs must be exactly 32 lowercase hexadecimal characters (no dashes). Strip dashes from UUIDs before calling "get_document" or "get_multiple_documents".