Back to MCP Servers

Fireflies MCP Server

Official Fireflies.ai MCP server for querying meeting transcripts, summaries, speaker analytics, soundbites, and channels via AI agents.

Collaboration by Fireflies.ai OAuth2 or Bearer Token (API Key) active
Overview

The Fireflies MCP Server is the official, Fireflies.ai-hosted Model Context Protocol endpoint that connects AI assistants to your meeting data. It exposes transcripts, meeting metadata, speaker information, AI summaries, soundbites, channels, and team analytics through a single remote server URL (https://api.fireflies.ai/mcp). Instead of copy-pasting transcript excerpts into a chat window, agents can search, retrieve, and act on meeting content directly inside Claude, ChatGPT, Cursor, Devin, and other MCP clients.

The server provides read tools for searching and listing transcripts, fetching full transcript details, pulling user profile and team analytics, listing channels, viewing active meetings, and retrieving soundbites. It also exposes mutation tools for sharing or revoking access to meetings, renaming meetings, moving them between channels, and creating soundbites. Authentication supports either OAuth (recommended, using your Fireflies/Google/Microsoft account, no API key required for ChatGPT and Claude connectors) or a Fireflies API key passed as a Bearer token for local clients like Claude Desktop.

Because it is provider-hosted, there is no GitHub repo to clone or process to self-manage. Setup is a single URL plus auth. Note that Fireflies' Zero-Day Retention policy does not extend to data processed through external MCP clients: those AI providers handle data according to their own terms.

Tools

Tool Description
fireflies_search_transcripts Search meeting transcripts by keyword, date range, or participant.
fireflies_get_transcript Retrieve full transcript details for a specific meeting.
fireflies_list_transcripts List meetings with filtering by organizer, participant, or date.
fireflies_get_user Fetch current user profile and settings.
fireflies_get_soundbites Fetch soundbites filtered by transcript, ownership, or team.
fireflies_list_channels List all accessible channels/folders.
fireflies_get_channel Get details of a specific channel by ID.
fireflies_get_active_meetings Retrieve currently in-progress meetings.
fireflies_get_analytics Obtain team and per-user meeting analytics.
fireflies_get_rule_executions View automation rule execution logs (Enterprise).
fireflies_share_meeting Share transcripts with up to 100 recipients, with optional expiry.
fireflies_revoke_meeting_access Revoke previously granted meeting access.
fireflies_update_meeting_title Modify a meeting transcript's title.
fireflies_move_meeting Relocate meetings to channels (up to 5 per request).
fireflies_create_soundbite Generate soundbite clips from meeting transcripts.
Setup Guide

Prerequisites

  • A Fireflies.ai account (Pro, Business, or Enterprise plan for full feature access)
  • An MCP-compatible client: Claude Desktop, Claude web, ChatGPT (Plus/Team/Enterprise), Cursor, VS Code, or Devin
  • Either a Fireflies API key (from Settings, Developer Settings) or OAuth via Google/Microsoft/Fireflies login

Server URL

https://api.fireflies.ai/mcp

Option 1: ChatGPT or Claude (OAuth, no API key)

In ChatGPT or Claude's connectors / MCP settings, add a new server with the URL https://api.fireflies.ai/mcp and complete the OAuth flow with your Fireflies account. No API key required.

Option 2: Claude Desktop with API key

Edit claude_desktop_config.json and add:

{
  "mcpServers": {
    "fireflies": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://api.fireflies.ai/mcp",
        "--header",
        "Authorization: Bearer YOUR_API_KEY_HERE"
      ]
    }
  }
}

Replace YOUR_API_KEY_HERE with the key from Fireflies Settings, Developer Settings. Restart Claude Desktop.

Option 3: Cursor / VS Code

Add the remote MCP server URL https://api.fireflies.ai/mcp through the MCP settings UI and authenticate via OAuth or supply the Bearer token header.

Use Cases
  • Summarize objections and competitor mentions across all sales demo transcripts from the last quarter.
  • Extract action items and decisions from a specific meeting and post them into a project tracker.
  • Search every customer call for mentions of a feature request and group them by account.
  • Pull team meeting analytics (talk time, meeting volume) for a weekly leadership report.
  • Auto-share a meeting transcript with a list of stakeholders and create a soundbite of the key moment.
Example Prompts
  • "Search my Fireflies transcripts from the last 30 days for any mention of 'pricing objection' and summarize the patterns."
  • "Get the full transcript for meeting abc123 and list every action item assigned to me."
  • "List all meetings organized by sarah@acme.com this month and show speaker talk time analytics."
  • "Find the soundbite where the customer described their onboarding problem and share that meeting with product@company.com."
  • "Move all of last week's discovery calls into the 'Q2 Pipeline' channel."
Pros
  • Official, Fireflies-hosted remote server: no self-hosting, no GitHub repo to maintain.
  • Broad tool coverage including both read (search, transcript, analytics, soundbites) and write (share, rename, move, create soundbite) operations.
  • Flexible auth: OAuth for ChatGPT and Claude connectors (no API key) or Bearer token for Claude Desktop and other local clients.
  • Works across major MCP clients: Claude, ChatGPT, Cursor, VS Code, Devin.
Limitations
  • Fireflies' Zero-Day Retention policy does not apply once data passes through an external MCP client; the AI vendor's terms govern that data.
  • Some tools (such as automation rule execution logs) require an Enterprise plan.
  • Public per-tool parameter documentation is still light compared to the GraphQL API reference.
Alternatives
  • Props-Labs/fireflies-mcp: community open-source MCP server wrapping the Fireflies GraphQL API.
  • Otter.ai or Grain integrations: similar meeting intelligence platforms, accessible via community MCP servers or REST APIs.
  • Direct use of the Fireflies GraphQL API (https://api.fireflies.ai/graphql) when you need lower-level control than MCP exposes.