AI Workforce MCP Server

A production-ready MCP (Model Context Protocol) server that provides Calendar, Email, Meeting Notetaker, and Weather tools. Deploy to Prefect Horizon for secure OAuth-authenticated access from any MCP client.

Features

• Calendar Management: List calendars, view events, create meetings with Google Meet

• Email Operations: List, analyze, and send emails

• Meeting Notetaker: Join meetings, record, transcribe, and generate AI summaries

• Weather Information: Get current weather for any city

• OAuth Authentication: Secure access via Prefect Horizon

Related MCP server: Atlas MCP Server

Available Tools

Deployment to Prefect Horizon

Prerequisites

1. A GitHub account

2. Your API keys ready:

   • Nylas API Key - Get from Nylas Dashboard

   • Nylas Grant ID - From your Nylas connected account

   • Weather API Key - Get from OpenWeatherMap

Step 1: Push to GitHub

# Initialize git repository (if not already)
cd mcp-server
git init

# Add all files
git add .

# Commit
git commit -m "Initial MCP server setup"

# Create a new repository on GitHub and push
git remote add origin https://github.com/YOUR_USERNAME/ai-workforce-mcp-server.git
git branch -M main
git push -u origin main

Step 2: Deploy to Prefect Horizon

1. Visit horizon.prefect.io

2. Sign in with your GitHub account

3. Connect Repository:

   • Grant Horizon access to your repositories

   • Select ai-workforce-mcp-server repository

4. Configure Server:

   • Server Name: ai-workforce (this becomes your URL: ai-workforce.fastmcp.app)

   • Description: "AI Workforce - Calendar, Email, and Meeting Tools"

   • Entrypoint: server.py:mcp

   • Authentication: Enable OAuth (recommended)

5. Add Environment Variables:

   • NYLAS_API_KEY: Your Nylas API key

   • NYLAS_GRANT_ID: Your Nylas grant ID

   • WEATHER_API_KEY: Your OpenWeatherMap API key

6. Click Deploy

Your server will be live at: https://ai-workforce.fastmcp.app/mcp

Connecting MCP Clients

Your MCP Server URL

After deployment, your server URL will be:

https://YOUR-SERVER-NAME.fastmcp.app/mcp

Claude Desktop Configuration

Add to your Claude Desktop config (claude_desktop_config.json):

{
  "mcpServers": {
    "ai-workforce": {
      "url": "https://ai-workforce.fastmcp.app/mcp",
      "transport": "streamable-http"
    }
  }
}

Claude Code CLI Configuration

# Add the remote MCP server
claude mcp add ai-workforce --url https://ai-workforce.fastmcp.app/mcp

Cursor IDE Configuration

Add to your Cursor settings:

{
  "mcp.servers": {
    "ai-workforce": {
      "url": "https://ai-workforce.fastmcp.app/mcp",
      "transport": "streamable-http"
    }
  }
}

Any MCP Client (Generic)

Use these connection details:

• URL: https://ai-workforce.fastmcp.app/mcp

• Transport: streamable-http

• Authentication: OAuth 2.1 (handled by Horizon)

OAuth Authentication Flow

When a new client connects:

1. Client requests connection to your MCP server URL

2. Horizon redirects to OAuth consent screen

3. User authenticates with their identity provider

4. Horizon issues access token

5. Client can now use all MCP tools

For automated/service accounts: Use Horizon API keys instead of interactive OAuth.

Local Development

Setup

# Clone the repository
git clone https://github.com/YOUR_USERNAME/ai-workforce-mcp-server.git
cd ai-workforce-mcp-server

# Create virtual environment
python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Copy environment file
cp .env.example .env

# Edit .env with your API keys

Run Locally

# Run with FastMCP CLI
fastmcp run server.py:mcp

# Or run directly
python server.py

Test with FastMCP Inspector

# Inspect tools before deployment
fastmcp inspect server.py:mcp

Testing Your Deployment

Using Horizon's Built-in Tools

1. Go to your server page on horizon.prefect.io

2. Click Inspector to see all tools and test them

3. Click ChatMCP for conversational testing

Using curl

# Get server status (after OAuth)
curl -X POST https://ai-workforce.fastmcp.app/mcp \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_TOKEN" \
  -d '{"method": "tools/call", "params": {"name": "get_server_status"}}'

Continuous Deployment

Horizon automatically:

• Redeploys when you push to main branch

• Creates preview deployments for pull requests

• Rolls back if deployment fails

Sharing Your Server

Public Access (with Authentication)

Anyone can connect to your server URL if they:

1. Have the URL: https://ai-workforce.fastmcp.app/mcp

2. Complete OAuth authentication

3. Are authorized in your Horizon settings

Managing Access

In Horizon dashboard:

• Organization members: Automatically authorized

• External users: Require explicit approval

• API Keys: For service accounts and automation

Troubleshooting

"No calendars found"

• Verify NYLAS_GRANT_ID is correct

• Ensure the grant has calendar permissions

"Authentication failed"

• Check NYLAS_API_KEY is valid

• Verify environment variables are set in Horizon

"Weather API error"

• Verify WEATHER_API_KEY is valid

• Check API quota limits

Server not responding

• Check Horizon dashboard for deployment logs

• Verify all required environment variables are set

Support

• Horizon Documentation: gofastmcp.com/deployment/prefect-horizon

• FastMCP Documentation: gofastmcp.com

• Nylas Documentation: developer.nylas.com

License

MIT License