Back to docs
Integrations · Granola MCP

Granola MCP bridge

Granola publishes its own MCP server at https://mcp.granola.ai/mcp. Add it to Claude Desktop, Cursor, or any MCP-compatible AI tool and your agent gets live access to your Granola meetings during a chat — without polling, copying, or going through mnueron.

When to use this instead of polling

Use MCP bridge when…

  • • You want live, mid-chat access to Granola meetings
  • • You don\'t need decisions / action items extracted into mnueron
  • • Your AI agent (not you) decides what to remember
  • • You already use mnueron MCP and want to add another source

Use mnueron polling when…

  • • You want every meeting in /dashboard/meetings
  • • You want auto-extracted decisions + action items
  • • You want one place to search across all sources
  • • You want shareable read-only meeting links

Most users benefit from both. Polling builds your searchable archive in mnueron; the MCP bridge gives your live AI sessions immediate access without round-tripping.

1

Add Granola MCP to your AI tool

The exact steps depend on which AI tool you use:

Claude Desktop

  1. Open Claude Desktop
  2. Settings → Connectors → Add custom connector
  3. URL: https://mcp.granola.ai/mcp
  4. Authenticate via the browser OAuth flow when prompted

Cursor

  1. Cursor → Settings → MCP
  2. Add MCP Server
  3. URL: https://mcp.granola.ai/mcp
  4. Transport: HTTP / SSE
  5. OAuth on first use

Claude Code (CLI)

  1. claude mcp add granola https://mcp.granola.ai/mcp
  2. Follow the browser OAuth prompt
  3. Restart Claude Code
2

Use it in chat

In Claude / Cursor / Cline, ask questions that need Granola context. Your agent will automatically call Granola MCP tools to read meeting notes.

You
What did we decide about pricing in yesterday\'s product review?
Agent
[searches Granola via MCP, finds the relevant note]
Yesterday\'s product review concluded: Pro at $19.99/mo as the primary tier, Team at $49/mo flat (not per-seat). Sarah owns the marketing update; Mike will adjust the Stripe prices today.
3

(Optional) Bridge MCP results into mnueron memory

If you also have mnueron MCP installed, your agent can persist what it found via memory_save. Just ask:

You: "Save that pricing decision to mnueron in namespace project:pricing."
Agent: [calls memory_save with the decision text] Saved memory id ab12cd34.

That memory is now searchable from every other AI tool wired to your mnueron account — Cursor recalls it, Claude Desktop recalls it, the dashboard shows it. The MCP bridge becomes the read side; mnueron becomes the persistent layer.

Two paths, one agent.

Mnueron polls Granola for your archive. Granola MCP gives your AI tool live access. Together they cover the entire lifecycle — automatic ingestion + on-demand recall.