What if Fireflies could answer questions like: "What were the main objections in this week's sales calls?" or "Create a summary of all product feedback from user interviews this month?"

…without you needing to switch platforms or copy-paste transcript excerpts?

The Fireflies MCP Server lets you turn what used to be 2–3 hours of manual transcript review into a 5-minute AI-powered analysis session — by bringing Fireflies meeting data into the AI agents and tools you already use.

How teams are using Fireflies MCP

Early users are integrating meeting intelligence into their AI workflows:

Sales teams are asking AI assistants to analyze pipeline conversations:

• "Compare the objections mentioned in calls with Enterprise prospects versus SMB prospects this quarter"

• "Generate a report on the most common questions asked during product demos."

Product managers are extracting research insights without manual review:

• "List all feature requests mentioned in user interviews, organized by frequency"

• "What integration requests came up in customer calls this month?"

• "Create a summary of usability feedback from the last 10 customer calls"

Customer success teams are monitoring account health across conversations:

• "Identify any red flags or satisfaction concerns mentioned in client calls this week"

• "Track progress on implementation issues discussed in recent support calls."

Team productivity and follow-ups:

• "What decisions were made in yesterday's product strategy meeting?"

• "Find meetings where we discussed [specific project or topic]"

The MCP server extends Fireflies' analytical capabilities into your existing AI workflow, letting you combine meeting insights with other data sources and analysis tools.

📋 How to set up Fireflies MCP

For ChatGPT 🆕

You can now connect Fireflies to ChatGPT using our OAuth integration—no API keys required.

Requirements:

• ChatGPT Plus, Team, or Enterprise account

• Access to ChatGPT Connectors feature

Setup steps:

1. Go to ChatGPT Settings → Connectors → Create

2. Enter connector details:

   • Name: Fireflies MCP (or your preferred name)

   • MCP Server URL: https://api.fireflies.ai/mcp

   • Authentication: OAuth

3. Check "I trust this application"

4. Click Create

5. Complete the OAuth authentication when prompted

Once connected, you can query your Fireflies data directly through ChatGPT's Deep Research feature.

For Claude 🆕

Fireflies is now officially listed in the Claude MCP directory! You can connect in just a few clicks.

Requirements:

• Paid Claude plan (Pro, Team, or Enterprise)

• Access to Claude Connectors feature

Setup steps:

1. Sign in to Claude ➡ Go to the Settings menu ➡ open the Connectors tab

2. Browse available connectors ➡ Click Browse Connectors to see the full list

3. Select Fireflies ➡ Find Fireflies in the list of MCPs and click on it

4. Connect and authorize ➡ Click Connect, then follow the prompts to authorize your OAuth connection

   

Alternative Setup Methods

If you prefer manual configuration or need to use the desktop app:

A. Use Google or Microsoft OAuth (Manual):

For Claude Web App:

1. Sign into your Claude and go to Connectors tab in Settings

2. Click Add custom connector

3. Name the connector and paste this URL: https://api.fireflies.ai/mcp and click Add

4. Once added, you will see this connector on the connectors tab by the name you entered above. In this case, Fireflies MCP. Click Connect

5. Once authenticated, Fireflies MCP will be available instantly

For Claude Desktop App:

Using the claude_desktop_config.json

Add the following config to your Claude Desktop Config file - You will be directed to the Google authentication page automatically

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

B. Use your Fireflies API key:

If you're not using Google OAuth, you can use the API key method on Claude Desktop

1. Go to Settings > Developer Settings and Copy your API key

   See how to get your API key →

2. Once you have the API key, follow the below steps:

3. Add this config to your claude.json file:

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

4. Finally, save the file and restart Claude. You should now be able to access all the Fireflies tools!

🔐 Security & Permissions

• Your API key is used to authenticate MCP access so store your key securely and do not share it publicly

• Your API key grants access to meeting transcripts and user data.

• Only the user's meetings are accessible via their own key

• Data access mirrors Fireflies' security standards (TLS encryption, etc.)

• If using scripts or local environments, consider storing the key using environment variables rather than hardcoding.

🛠 Tools available via MCP

🔧 Troubleshooting

Common Issues:

• Authentication Error: Double-check your Fireflies API key is valid and hasn't expired.

• No Transcripts Found: Ensure your Fireflies account has processed meetings with transcripts available.

• Connection Issues: Restart your integration tool (e.g., Claude, ChatGPT) and verify your internet connection.

🧠 FAQs

Is this a replacement for the Fireflies UI?

No. MCP extends your meeting data to external AI tools. The Fireflies app still offers deeper UI-based insights, integrations, and collaboration tools.

Does this work with ChatGPT or other models?

Yes! ChatGPT now has native support via our OAuth connector. Claude also works via their desktop or web integration options. Other tools could support it if they allow configuring MCP-compatible context servers or custom API routing.

Can I use this for my whole team?

Each user must use their own API key for now. Team-level or workspace-wide support is in development.

Can I filter by topic, date, or speaker?

Yes — you can apply filters while querying for meeting transcripts with prompt engineering

Do I still need an API key to use MCP?

Not if you're signing in with Google or using the ChatGPT OAuth connector. Our OAuth-based flows let users connect instantly—no config files or API keys needed.