> This page is for version v8.3.0 (default).
> For other versions, use one of these documentation indexes:
> - v8.3.0 (default): https://docs.workflows.com/v-8-3-0/llms.txt

> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.workflows.com/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.workflows.com/_mcp/server.

# Fireflies Integration Setup

> **Access Meeting Transcripts**
>
> Connect Fireflies to pull meeting transcripts directly into your AI workflows. Search across your entire transcript library and create minds from team meetings and client calls.

## What You'll Get

When you connect Fireflies to your MCP server, you unlock **3 additional tools**:

* **fireflies_list_transcripts** - List meeting transcripts with date filtering
* **fireflies_get_transcript** - Get detailed transcripts with speaker analytics
* **fireflies_search_transcripts** - Search transcripts by keyword across your entire library

These tools work alongside the 8 core Mind Reasoner tools, giving you seamless access to your meeting data.

## Important: Speaker Email Limitation

> **Speaker Email Attribution**
>
> **Fireflies does not directly provide speaker email addresses in transcript data.** Instead, the MCP server correlates speaker names with meeting attendee emails via name matching.
>
> **What this means:**
>
> * Speaker emails are inferred, not guaranteed
> * Name matching may be imperfect (e.g., "John" vs "John Smith")
> * Some speakers may not have email attribution if names don't match attendees
>
> **If you need reliable speaker emails:** Consider using Gong or Fathom, which provide direct speaker email attribution in their APIs.

## Choose Your Integration Approach

You have two options for working with Fireflies data:

### Option 1: MCP Server Integration (Recommended)

**Best for:** AI-driven workflows, natural language access, no coding

**Benefits:**

* Access Fireflies transcripts through simple AI commands
* Powerful keyword search across all transcripts
* Automatic authentication and error handling
* Seamless integration with Mind Reasoner mind creation
* Works in Claude Desktop, Claude Code, and other MCP clients

**Setup:** Follow the steps below to add Fireflies credentials to your MCP server

### Option 2: Direct Fireflies GraphQL API Integration

**Best for:** Custom applications, specialized workflows, full API control

**Benefits:**

* Complete control over GraphQL queries and data processing
* Build custom integrations with other services
* Implement specialized business logic
* Access to all Fireflies data fields

**Resources:**

* [Fireflies GraphQL API Documentation](https://docs.fireflies.ai/graphql-api)
* [Fireflies API Playground](https://app.fireflies.ai/integrations/custom/api)
* [Mind Reasoner REST API](/api-reference) - Combine with Fireflies data as needed

---

## MCP Server Integration Setup

#### Get Your Fireflies API Key

**1. Log in to Fireflies**

Navigate to [app.fireflies.ai](https://app.fireflies.ai)

**2. Access Integration Settings**

Click your profile icon → **Settings** → **Integrations** → **API Key**

Or visit directly: [app.fireflies.ai/integrations/custom/api](https://app.fireflies.ai/integrations/custom/api)

**3. Generate API Key**

Click **"Generate API Key"** or copy your existing key

The API key will be a long alphanumeric string.

> **API Key Management**
>
> You can view your Fireflies API key again in settings if needed. However, for security, treat it as a secret and store it securely.

#### Configure Your MCP Server

The configuration method depends on how you're running the MCP server:

### For Local/Stdio Mode (Claude Desktop, Claude Code CLI)

Add this environment variable to your configuration:

**`Environment Variables (.env)`**

```bash title="Environment Variables (.env)"
# Required - Mind Reasoner
MIND_REASONER_API_KEY=your_mind_reasoner_key

# Optional - Fireflies Integration
FIREFLIES_API_KEY={{FIREFLIES_API_KEY}}
```

### For Hosted/HTTP Mode (Serverless, Custom Clients)

Pass credentials as HTTP headers with each request:

**`HTTP Headers`**

```bash title="HTTP Headers"
# Required
Authorization: Bearer YOUR_MIND_REASONER_KEY

# Optional - Fireflies
X-Fireflies-Api-Key: {{FIREFLIES_API_KEY}}
```

> **Per-Request Authentication (HTTP Mode)**
>
> In HTTP mode, you can provide different Fireflies credentials for each request. This is useful for multi-tenant applications or supporting multiple Fireflies accounts.

#### Restart Your MCP Client

If using Claude Desktop or Claude Code:

```bash
# Restart Claude Desktop or Claude Code
# The Fireflies tools will be available automatically
```

No restart needed for HTTP/serverless mode—changes take effect immediately.

#### Verify the Connection

Test that Fireflies tools are available:

**Ask your AI assistant:**

```
"What Fireflies tools do you have access to?"
```

**Expected response:**

* fireflies_list_transcripts
* fireflies_get_transcript
* fireflies_search_transcripts

**Try a test query:**

```
"Show me my last 5 Fireflies transcripts from this week"
```

If you see transcript data, you're all set!

> **Integration Active**
>
> Fireflies is now connected to your MCP server. You can access meeting transcripts and search across your entire library through natural language commands.

---

## Authentication Details

### How Fireflies Authentication Works

Fireflies uses **Bearer Token authentication** with their GraphQL API:

1. Your API key is sent as a Bearer token in the `Authorization` header
2. The MCP server constructs GraphQL queries to fetch transcript data
3. Fireflies validates the token and returns data accessible to your account

The MCP server handles all GraphQL query construction, so you don't need to know GraphQL to use these tools.

### Security Best Practices

> **Protect Your Fireflies API Key**
>
> Your Fireflies API key provides access to all meeting transcripts in your account.
>
> **Security checklist:**
>
> * Never share your API key in messages or screenshots
> * Don't commit API keys to version control (use .env files, add to .gitignore)
> * Use environment variables for local development
> * Regenerate your key periodically from Fireflies settings
> * If compromised, immediately revoke and generate a new key
> * Monitor API usage in Fireflies settings regularly

---

## Troubleshooting

### "Fireflies tools not available"

**Possible causes:**

* Credentials not configured
* Environment variables not loaded
* MCP client not restarted after configuration

**Solutions:**

1. Verify environment variable is set correctly
2. Check for typos in variable name (`FIREFLIES_API_KEY`)
3. Restart your MCP client (Claude Desktop, Claude Code)
4. Ask: "What MCP tools are available?" to see current tool list

### "Authentication failed" or 401 errors

**Possible causes:**

* Incorrect API key
* API key revoked or deleted
* Fireflies account suspended

**Solutions:**

1. Verify API key matches exactly what's shown in Fireflies settings
2. Check for extra spaces or line breaks in the key
3. Generate a new API key in Fireflies settings
4. Verify your Fireflies account is active and has API access

### "Rate limit exceeded" or 429 errors

**Cause:** Too many API requests to Fireflies in a short time period

**Solutions:**

1. Wait before making additional requests
2. Reduce the frequency of transcript queries
3. Use date filtering to narrow search results
4. Contact Fireflies support to discuss rate limits

### No transcripts returned from fireflies_list_transcripts

**Possible causes:**

* No meetings in the specified date range
* User permissions don't include transcript access
* Fireflies account has no recorded meetings

**Solutions:**

1. Try a wider date range or remove date filter
2. Verify your Fireflies account has recorded meetings
3. Check that meetings are fully processed (not still transcribing)
4. Ensure you have access to the transcripts (not restricted)

### Speaker emails missing or incorrect

**Cause:** Fireflies doesn't directly provide speaker emails; the MCP server infers them via name matching

**Understanding the limitation:**

* Fireflies provides speaker names but not emails in transcript data
* The MCP server correlates speaker names with meeting attendee emails
* Name matching is best-effort (handles variations, but not perfect)
* Some speakers may not match any attendee

**Solutions:**

1. Verify meeting attendee list includes the speakers
2. Check that speaker names match attendee names reasonably well
3. For critical use cases requiring accurate emails, use Gong or Fathom instead
4. Manually verify and correct speaker attribution when creating minds

### Search returns too many/few results

**Cause:** Keyword search across all transcript text can be broad or narrow

**Solutions:**

1. Use more specific keywords to narrow results
2. Use date filtering in combination with keyword search
3. Try alternative keywords or phrases
4. Use wildcard or partial matching in search terms

---

## Example Workflows

Once Fireflies is connected, try these workflows:

### List Recent Transcripts

```
"Show me my last 10 Fireflies transcripts from this month"
```

AI uses: `fireflies_list_transcripts` with date filtering

### Search for Specific Topics

```
"Search my Fireflies transcripts for mentions of 'product roadmap'"
```

AI uses: `fireflies_search_transcripts` with keyword "product roadmap"

### Create Mind from Specific Meeting

```
"Create a mind from my Fireflies meeting about Q4 planning"
```

AI workflow:

1. Searches transcripts for "Q4 planning"
2. Retrieves full transcript with `fireflies_get_transcript`
3. Creates a new mind with `create_mind`
4. Uploads transcript data
5. Creates snapshot and waits for training

### Analyze Customer Feedback

```
"Search all my customer meetings for mentions of 'pricing'
 and create a summary mind of customer sentiment"
```

AI orchestrates:

1. Searches transcripts for "pricing" keyword
2. Gets full transcripts for matching meetings
3. Filters to customer-facing participants
4. Creates consolidated mind from all relevant segments
5. Trains on customer pricing discussions

### Weekly Team Meeting Analysis

```
"Find all transcripts from my weekly team standup meetings
 in the last month and create minds for each team member"
```

AI workflow:

1. Searches for "standup" or "weekly team" in transcripts
2. Filters to last month
3. Gets full transcripts for all matching meetings
4. Identifies unique team members across meetings
5. Creates minds for each person with their segments
6. Trains all minds in parallel

---

## Fireflies Unique Features

### Advanced Search

Fireflies provides powerful transcript search that's unique among the integrations:

* **Keyword Search:** Find specific terms across all transcripts
* **Phrase Matching:** Search for exact phrases
* **Date Filtering:** Combine keyword search with date ranges
* **Speaker Analytics:** See which speakers mentioned specific topics

**Example:**

```
"Search my Fireflies transcripts from Q3 for mentions
 of 'customer retention' and show me which speakers
 discussed it most"
```

### GraphQL Backend

Unlike Gong (REST) and Fathom (REST), Fireflies uses GraphQL:

* **Efficient Data Fetching:** Request exactly the fields you need
* **Flexible Queries:** Combine multiple data requirements in one request
* **Rich Data Model:** Access nested data structures easily

The MCP server handles all GraphQL complexity automatically.

---

## Data Access & Privacy

### What Data You Can Access

Fireflies provides transcript data through the API:

* **Meeting Metadata:** Title, date, duration, participants (names)
* **Transcripts:** Full text with speaker attribution (names only)
* **Speaker Analytics:** Talk time, word count, speaking patterns
* **Search Results:** Keyword matches across transcript text
* **Attendees:** Meeting participant names (emails inferred via matching)

### Data Privacy

When you connect Fireflies:

* Only transcripts you have access to are retrievable
* Transcripts are fetched in real-time from Fireflies' GraphQL API
* Data is transmitted securely to Mind Reasoner for processing
* The MCP server does not store transcript data
* All data handling follows Mind Reasoner's privacy policies

---

## Next Steps

🔧

### Explore Fireflies Tools

Learn about all 3 Fireflies tools, their parameters, and detailed usage examples.

[Fireflies Tools Reference →](/mcp-integrations-fireflies)

📞

### Add More Integrations

Connect Gong or Fathom for access to sales calls and meeting transcripts with full speaker email attribution.

[All Integrations →](/mcp-integrations)

💡

### Complete Workflows

See end-to-end examples of creating minds from Fireflies data.

[Quickstart Guide →](/mcp-quickstart)

## Official Documentation

For direct GraphQL API integration or advanced use cases:

* [Fireflies GraphQL API Documentation](https://docs.fireflies.ai/graphql-api) - Complete API guide
* [Fireflies API Playground](https://app.fireflies.ai/integrations/custom/api) - Test queries interactively
* [Fireflies Help Center](https://help.fireflies.ai/) - General Fireflies support