Assistant and MCP Integration

The LUMA assistant is available through the Model Context Protocol (MCP), letting you query your financial data from AI tools like Claude Desktop, Cursor, ChatGPT, and Raycast.

#What is MCP?

Model Context Protocol (MCP) is an open standard for connecting AI assistants to external data sources. Instead of copying and pasting data, MCP lets AI tools query your LUMA data directly and securely.

#Supported tools

LUMA MCP works with:

• Claude Desktop - Anthropic's desktop app
• Cursor - AI-powered code editor
• ChatGPT - Via plugins/actions
• Raycast - Mac productivity launcher
• Any MCP-compatible client

#Getting started

#Step 1: Get your API key

1. Go to Settings → Developer
2. Click Create API key
3. Give it a descriptive name (e.g., "Claude Desktop")
4. Copy the key immediately (you won't see it again)

#Step 2: Configure your client

Choose your tool below for specific setup instructions.

---

#Claude Desktop setup

Add LUMA to your Claude Desktop configuration.

#Find your config file

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

#Add LUMA configuration

Add or update the mcpServers section:

{
  "mcpServers": {
    "luma": {
      "command": "npx",
      "args": ["-y", "@luma-ai/mcp"],
      "env": {
        "LUMA_API_KEY": "your-api-key-here"
      }
    }
  }
}

#Restart Claude Desktop

After saving, restart Claude Desktop. You should see LUMA in your available tools.

#Test it

Ask Claude something like:

• "What was my revenue last month?"
• "Show me unpaid invoices"
• "What's my current runway?"

---

#Cursor setup

Use LUMA data while coding in Cursor.

#Open Cursor settings

1. Open Cursor
2. Go to Settings (Cmd/Ctrl + ,)
3. Find the MCP configuration section

#Add LUMA

Add to your MCP servers:

{
  "luma": {
    "command": "npx",
    "args": ["-y", "@luma-ai/mcp"],
    "env": {
      "LUMA_API_KEY": "your-api-key-here"
    }
  }
}

#Use in conversations

In Cursor's AI chat, you can now ask about your financial data while working on your projects.

---

#ChatGPT setup

Use LUMA with ChatGPT via custom GPTs or plugins.

#Custom GPT approach

1. Create a Custom GPT in ChatGPT
2. Add actions that call the LUMA API
3. Configure authentication with your API key

#Configuration

Use the LUMA API endpoints with your API key in the Authorization header:

Authorization: Bearer your-api-key-here

See the API reference for available endpoints.

---

#Raycast setup

Access LUMA from Raycast on Mac.

#Install the extension

1. Open Raycast
2. Search for "LUMA" in the Store
3. Install the extension

#Configure

1. Open Raycast preferences
2. Find the LUMA extension
3. Enter your API key

#Use it

Trigger Raycast and use LUMA commands to query your financial data.

---

#Available tools

The MCP integration provides access to all your LUMA data:

#Transactions

Example queries:

• "Show transactions from last month"
• "What did I spend on software this quarter?"
• "Find transactions over $1,000"

#Invoices

Example queries:

• "Show unpaid invoices"
• "What invoices are overdue?"
• "Total outstanding amount"

#Customers

Example queries:

• "List all my customers"
• "Show details for Acme Corp"

#Time tracking

Example queries:

• "How many hours did I track this week?"
• "Show time by project this month"
• "What's unbilled for Project X?"

#Reports

Example queries:

• "What's my revenue this quarter?"
• "Calculate my runway"
• "Show spending by category"

#Bank accounts

Example queries:

• "What's my current cash balance?"
• "Show all connected accounts"

#Documents

#Inbox

#Search

Example queries:

• "Search for Amazon"
• "Find anything related to Stripe"

#Team

---

#Example conversations

#Financial overview

You: What's my financial health looking like?

Assistant: Based on your LUMA data:

• Cash balance: $45,230
• Monthly burn rate: $8,500
• Runway: 5.3 months
• Revenue this month: $12,400
• Profit margin: 31%

#Invoice management

You: Do I have any overdue invoices?

Assistant: You have 2 overdue invoices:

1. INV-0042 to Acme Corp - $3,500 (15 days overdue)
2. INV-0039 to Beta Inc - $1,200 (8 days overdue)

Total overdue: $4,700

#Expense analysis

You: What are my top expenses this quarter?

Assistant: Your top expense categories this quarter:

1. Salary: $24,000
2. Software: $3,200
3. Marketing: $2,800
4. Rent: $2,400
5. Professional Services: $1,500

---

#Security

#API key permissions

• API keys have read-only access by default
• Cannot create, modify, or delete data
• Cannot access sensitive credentials or payment information

#Managing keys

1. Go to Settings → Developer
2. View all active API keys
3. See last used date for each key
4. Revoke keys you no longer need

#Best practices

• Create separate keys for different tools
• Use descriptive names for keys
• Revoke unused keys
• Don't share keys publicly

---

#Troubleshooting

#"Tool not found" errors

• Ensure you've restarted your client after configuration
• Check that the API key is valid
• Verify the configuration syntax

#No data returned

• Check your API key has access to the team
• Verify transactions/invoices exist for the queried period
• Try a broader query

#Connection issues

• Ensure you have internet access
• Check that npx is available in your PATH
• Try running npx @luma-ai/mcp manually to test

---

#Tips for best results

1. Be specific: "Revenue last month" works better than "How am I doing?"
2. Include dates: "Expenses in Q4 2024" helps narrow results
3. Use follow-ups: The assistant remembers context in a conversation
4. Try different phrasings: If one query doesn't work, rephrase it

Learn about the in-app assistant → API reference →