Fathom MCP Server
Enables querying meeting transcripts, summaries, and action items from Fathom directly in your MCP-compatible AI assistant, eliminating context switching while coding.
README
Fathom MCP Server
Stop switching between your IDE and Fathom's web UI. Query your meeting transcripts, summaries, and action items directly from Cursor, Claude Desktop, or any MCP-compatible AI assistant.
The actual use case: You're coding and remember someone mentioned a specific UI change in last week's design review. Instead of opening Fathom, finding the meeting, and scrubbing through the recording, just ask your AI: "What did Sarah say about the dashboard layout in last week's design meeting?" and implement it immediately.
Built for developers who want meeting context without context switching.
Features
- List Meetings: Search and filter meetings by date, teams, participants, or domains
- Get Summaries: Retrieve AI-generated summaries for specific recordings
- Get Transcripts: Access full meeting transcripts with speaker attribution
- Team Management: List teams and team members in your organization
- Action Items: Extract action items from meetings
- CRM Integration: Include CRM matches in meeting data
Quick Setup
Prerequisites
- A Fathom account with API access
- Your Fathom API key (Get it here)
Configuration
Add this to your MCP client configuration:
For Cursor
Add to ~/.cursor/mcp.json (or create it if it doesn't exist):
{
"mcpServers": {
"fathom": {
"transport": "streamable_http",
"url": "https://fathommcp-production.up.railway.app/mcp",
"headers": {
"x-api-key": "YOUR_FATHOM_API_KEY"
}
}
}
}
For Claude Desktop
Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):
{
"mcpServers": {
"fathom": {
"transport": "streamable_http",
"url": "https://fathommcp-production.up.railway.app/mcp",
"headers": {
"x-api-key": "YOUR_FATHOM_API_KEY"
}
}
}
}
For Other MCP-Compatible Clients
Use the same configuration format with:
- Transport:
streamable_http - URL:
https://fathommcp-production.up.railway.app/mcp - Header:
x-api-keywith your Fathom API key
Getting Your Fathom API Key
- Log in to Fathom
- Go to Settings → Integrations → API
- Generate or copy your API key
- Replace
YOUR_FATHOM_API_KEYin the configuration above
Real-World Usage
This isn't just another MCP server - it puts your meeting context directly into your coding workflow. Here's how you actually use it:
In Cursor (while coding)
Implement based on meeting feedback:
- "Pull up the transcript from yesterday's product review with Sarah - what UI changes did she request for the dashboard?"
- "What did the design team say about the button colors in last Friday's meeting? Update the theme file based on that"
- "Check what Mark mentioned about the API response format in this week's standup and refactor accordingly"
Debug with context:
- "Find the meeting where we discussed the authentication bug - what was the root cause they identified?"
- "What did the customer say was broken in their demo call? Let me fix that exact flow"
Build features from discussions:
- "Get the specs for the export feature from the planning meeting with engineering"
- "What were the exact requirements the client mentioned for the search functionality? Implement that"
Reference technical decisions:
- "Why did we decide to use Redis over PostgreSQL for caching? Pull that from the architecture meeting"
- "What performance targets did we agree on in the infrastructure call?"
In Claude Desktop
Meeting prep & follow-up:
- "Summarize all meetings with Acme Corp from last month"
- "What action items did I commit to in this week's meetings?"
- "Pull the transcript from the sales call - I need to draft a follow-up email"
Cross-reference discussions:
- "Find all meetings where we discussed the pricing model"
- "Show me what different teams said about the mobile app redesign"
Available Tools
The MCP server provides these tools:
list_meetings- Search and filter meetings with optional transcript/summary inclusionget_summary- Get AI summary for a specific recordingget_transcript- Get full transcript for a specific recordinglist_teams- List all teams in your organizationlist_team_members- List members of your teams
Self-Hosting
To run your own instance:
Local Development
# Clone the repository
git clone <your-repo-url>
cd fathom_mcp
# Install dependencies
pip install -r requirements.txt
# Run the server
python fathom_mcp.py
Deploy to Railway
- Fork this repository
- Connect to Railway
- Deploy from your GitHub repository
- Railway will automatically detect and deploy the FastMCP server
Why This Is Actually Useful
Stop context switching. Instead of pausing your code to dig through Fathom's web UI, just ask while you're in your IDE:
- Code from memory, not notes: "What did the PM say about error handling?" - implement it right then and there
- Ship what was actually requested: No more building features based on vague Slack messages. Get the exact words from the meeting
- Debug faster: "Pull the bug description from the customer call" - fix what they actually reported, not what got lost in translation
- No meeting notes FOMO: Missed a standup? Get the key points without scrubbing through a recording
- Build better demos: "What questions did the client ask last time?" - address them proactively in your next version
- Onboard faster: New to a project? Search all past meetings about it and get up to speed
- Write better PRs: Reference the exact meeting discussion when explaining why you made specific technical decisions
API Documentation
This server uses the Fathom External API v1. All tools support the full range of filters and options available in the Fathom API.
Support
Need help setting it up? Feel free to reach out or open an issue.
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。