note1

note1 MCP Server

Upload recordings and work with your meeting notes, transcripts, and action items from Claude, ChatGPT, Cursor, or any MCP client.

note1 is an AI meeting notetaker that joins your calls, records them, and turns every meeting into a searchable summary with action items. This MCP (Model Context Protocol) server connects AI tools directly to that meeting data — search across conversations with cited sources, read who spoke how much and what was asked, export summaries and speaker-attributed transcripts, list, create and complete action items (and close their linked tickets in GitHub, Linear, Jira and more), manage topic trackers, connect and configure team integrations, read and update your meeting notes, check which calendar events are being recorded, and schedule or manage recordings, all from a conversation.

Every tool acts with the authenticated user's own permissions: private meetings stay private, and results match exactly what the user sees in the note1 dashboard.

• Server URL: https://api.note1.ai/mcp (Streamable HTTP)
• Registry: ai.note1/mcp in the official MCP Registry
• Docs: note1.ai/docs/mcp-server
• Auth: OAuth 2.1 (sign in to note1 in your browser) or a personal access token

Connect without this package

Clients that support remote MCP servers can connect to https://api.note1.ai/mcp directly and sign in to note1 in the browser. You need neither this package nor a token.

• Claude and ChatGPT: add a custom connector with the URL https://api.note1.ai/mcp, then sign in to note1 and approve access. Custom connectors depend on your plan.
• Claude Code: run claude mcp add --transport http note1 https://api.note1.ai/mcp, then run /mcp and sign in.
• Cursor and VS Code: add the URL as a remote MCP server and sign in when asked.
• Slack (Slackbot): workspaces connected to note1 get zero-config access. Open a DM with Slackbot, choose Apps, and add note1. Slack identifies you by your workspace email.

Setup with this package

You need a note1 account (note1.ai). Every client below runs the same command, npx -y @note1ai/mcp. Requires Node.js 18 or newer.

The first time the server starts, the package opens note1 in your browser. Sign in and approve access. The sign-in is saved in ~/.mcp-auth on your computer, so you do not repeat it every time.

Claude Desktop

Add to claude_desktop_config.json:

{

"mcpServers": { "note1": { "command": "npx", "args": ["-y", "@note1ai/mcp"] } } }

Cursor

Add to ~/.cursor/mcp.json (or .cursor/mcp.json in a project). Cursor can also connect directly; use this package if Cursor's own sign-in fails.

{

"mcpServers": { "note1": { "command": "npx", "args": ["-y", "@note1ai/mcp"] } } }

VS Code

Add to .vscode/mcp.json:

{

"servers": { "note1": { "type": "stdio", "command": "npx", "args": ["-y", "@note1ai/mcp"] } } }

Claude Code

claude mcp add note1 -- npx -y @note1ai/mcp

Any other MCP client

Configure a stdio server with command npx and arguments -y @note1ai/mcp.

Use a token instead

Browser sign-in needs a browser on the same computer as the MCP client. Over SSH, in CI or in a container, or to skip the sign-in, set NOTE1_API_TOKEN to a personal access token (starts with n1_). Create one at note1.ai/dashboard/account/tokens. It is shown once, so copy it.

{

"mcpServers": { "note1": { "command": "npx", "args": ["-y", "@note1ai/mcp"], "env": { "NOTE1_API_TOKEN": "n1_YOURTOKEN" } } } }

With Claude Code:

claude mcp add note1 --env NOTE1_API_TOKEN=n1_YOURTOKEN -- npx -y @note1ai/mcp

Configuration

The token is passed to mcp-remote through its environment, not on the command line, so it does not show up in process lists. Extra arguments given to the command are forwarded to mcp-remote. The server URL is fixed to https://api.note1.ai/mcp.

Tools

Upload an existing recording

Recording uploads require meetings:write and the user's approval. One private meeting holds one immutable original audio/video recording; uploads never launch a meeting bot. Use note1_get_recording_upload_options first to check the current workspace limit (initially 2,000,000,000 bytes).

For direct transfer, your assistant needs local-file access and an HTTP client. Call note1_upload_recording with metadata only:

{

"requestId": "fe926ddd-03bf-4917-8f1b-b44d3f290049", "title": "Customer interview", "file": { "name": "interview.wav", "size": 123456, "mimeType": "audio/wav" }, "language": "en", "style": "interview" }

The tool reserves a meeting and returns IDs plus transfer instructions. Outside MCP, POST each exact file slice using the returned offset, size, and integer chunk index in urlTemplate. Send the returned Content-Type: application/octet-stream and X-Note1-Upload-Token headers. Never pass bytes, base64, local paths, or account-wide credentials through upload tool arguments; never put the upload token in a URL.

Check acknowledged chunks with note1_get_recording_upload after a lost response. Retry initialization with the same requestId and frozen file metadata to recover or renew the upload permission. Identical chunk retries do not charge twice. After all bytes are acknowledged, call note1_complete_recording_upload with uploadId and fileId; processing may still be running, so use note1_get_meeting for results. Keep the same resolved teamId throughout.

If your assistant cannot transfer local files, use the browser link from note1_get_recording_upload_options before reserving. Sign in, verify or explicitly switch to the requested workspace, then choose and submit a file. If login sends you to the dashboard, reopen the assistant's link. Opening the page or form creates no meeting. The browser starts its own upload, not cross-tab resume; continue an existing direct reservation or confirm cancellation before starting over. Completion opens the meeting page, without an automatic callback into the chat.

Initialization reserves one meeting allowance; normal storage, transfer, and processing accounting applies. Cancellation does not refund meeting allowance or transfer usage. If completion already won, check status and open the existing meeting rather than claiming