Meetric MCP

Connect your AI assistant directly to your Meetric meeting data using the Model Context Protocol.

Beta. This connector is read-only, the tool set may change, and rate limits apply. Please don't depend on it for production-critical automation yet.

Overview

MCP is an open standard that lets AI tools (like Claude, Cursor, Codex, and other MCP-compatible assistants) securely call external tools. Meetric hosts an MCP server so your assistant can read your meeting data — conversations, transcripts, summaries, insights, users, teams, and departments — through your signed-in user permissions. You connect by logging in with your normal account; there are no tokens to copy.

http://127.0.0.1:36569/mcp JSON-RPC 2.0 over HTTPS

Switched off by default. MCP is switched off for every account by default. An account owner or admin switches it on in the app under Settings → Account → AI assistant connection (MCP). If your account comes from a partner, the partner has to allow MCP first.

How it works

1. User sign-in: OAuth-capable MCP clients (including Notion) dynamically register, then open a Meetric login screen. Use your normal password, SSO, and MFA flow.
2. Your permissions apply: the connection is tied to your user and account — the MCP can only ever see what you can see in the app.
3. Your assistant calls tools/list to discover tools, then tools/call to fetch data. Every call is validated against the Meetric API and scoped to your account.

Authentication

Meetric MCP uses user sign-in only: OAuth + PKCE with Dynamic Client Registration, issuing a mcp_at_… token tied to your user and account. Invalid or expired tokens are rejected with 401. Account API tokens (mt_…) work on the REST API, but are not a way into the MCP — every MCP read belongs to a signed-in user.

The client opens Meetric login, including password, SSO, and MFA when required. Data access follows the signed-in user's normal Meetric permissions, and changes to your role, account, or licence apply on the next call — no reconnecting needed.

Token lifetime. An access token (mcp_at_…) is valid for 60 minutes. AI clients renew it on their own with the refresh token, which is valid for 30 days and renewed on every use, so you stay signed in for as long as you keep using the connection. If you build your own integration, it must implement the OAuth refresh_token grant: a copied access token stops working after an hour. For automation that runs without a signed-in user, use the Meetric REST API with an API token instead.

Connect your assistant

Claude Desktop

Claude Desktop runs MCP servers locally, so bridge the hosted endpoint with mcp-remote (it handles the sign-in for you). Add this to claude_desktop_config.json (Settings → Developer → Edit Config):

{
  "mcpServers": {
    "Meetric": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "http://127.0.0.1:36569/mcp"]
    }
  }
}

Cursor

Cursor supports remote MCP servers directly. Add this to ~/.cursor/mcp.json (or Settings → MCP) and sign in when prompted:

{
  "mcpServers": {
    "Meetric": {
      "url": "http://127.0.0.1:36569/mcp"
    }
  }
}

Codex (OpenAI Codex CLI)

Add this to ~/.codex/config.toml:

[mcp_servers.meetric]
command = "npx"
args = ["-y", "mcp-remote", "http://127.0.0.1:36569/mcp"]

Notion & other MCP-compatible tools

For any assistant that supports a remote / custom MCP connector, add the endpoint — the connector opens the Meetric sign-in when you connect:

{
  "mcpServers": {
    "Meetric": {
      "url": "http://127.0.0.1:36569/mcp"
    }
  }
}

If the tool only supports local (stdio) MCP servers, use the mcp-remote bridge shown in the Claude example.

Verify it works

List the available tools from a terminal (replace the token with the mcp_at_… access token your client obtained by signing in):

curl -s http://127.0.0.1:36569/mcp \
  -H "Authorization: Bearer YOUR_MCP_ACCESS_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

A 401 answer without a token is expected — it means the endpoint is alive and asking for sign-in. Your assistant handles that automatically.

Available tools

Tool	Scope	Description
account_get	—	Get basic account and branding information for the connected account.
users_list	users	List active users in the account. Requires scope: users.
teams_list	teams	List teams and members in the account. Requires scope: teams.
departments_list	departments	List enabled departments in the account. Requires scope: departments.
conversations_list	conversations	List recent conversations you can see, with the standard filters.
conversations_get	conversations	Get one conversation by ID. Requires scope: conversations.
conversations_transcript	transcripts	Get transcript for one conversation. Requires scope: transcripts.
conversations_insights	insights	Get insights/topics for one conversation. Requires scope: insights.
conversations_summary	conversations	Get summary for one conversation. Requires scope: conversations.
topics_list	topics	List topics (KPI-grouped insights) you can access, aggregated across your visible conversations.
topics_drilldown	topics	Drill into a topic to get its underlying insights with source conversation references and transcript timestamps. Provide kpi_label and/or insight_type from topics_list.
search	search	Search transcripts and insights (semantic + exact keyword) for words or phrases across the conversations you can access. Returns hits with conversation references + timestamps for transcript drilldown.

Limits & behavior

• Read-only. No tool writes or modifies Meetric data.
• Account-scoped. A connection only ever sees the data of its own account, filtered by the signed-in user's permissions — transcripts, private meetings, and licences follow the same rules as in the app.
• Pagination. List results are capped (limit is clamped to 25); use the returned cursor for more.
• Request size. Requests are limited to 64 KB.

Important: GDPR & compliance boundaries. While Meetric ensures secure data transit and protocol compliance within the MCP environment, you are solely responsible for ensuring that all downstream applications, custom connected servers, and external data storage locations comply with your organization's GDPR policies and local privacy regulations. Meetric does not monitor or manage compliance rules applied outside of the Meetric platform.

Support

Questions or feedback? Email [email protected].

Meetric MCP — Beta · Endpoint http://127.0.0.1:36569/mcp