Ask AI About Your Cloud Costs: Using the Vantage MCP Server with Claude, Cursor, and Goose
Tired of squinting at spreadsheets to understand your cloud spend? The Vantage MCP Server brings the power of natural language to your cost analysis, letting you ask simple questions and get clear answers.
This article dives into how the Vantage MCP Server works and how you can integrate it with popular AI tools like Claude, Cursor, and Goose to unlock a new level of cloud cost visibility.
What is the Vantage MCP Server?
The Vantage MCP Server is an open-source tool that acts as a bridge between AI assistants and the Vantage API. Written in Go, it allows you to query Vantage's cost data using natural language through MCP (Model Context Protocol) clients. Imagine asking "What were my AWS costs last month?" and getting an immediate, accurate answer.
- Open-source: Fully transparent and customizable.
- Natural language queries: Makes cost analysis accessible to everyone.
- Go-based: Efficient and reliable.
Tools Available within the MCP Server
The Vantage MCP Server exposes several useful tools that AI clients can invoke:
get-costs
: Filters costs by date and Vantage Query Language (VQL).list-cost-reports
: Lists all available cost reports.list-cost-integrations
: Shows available cost provider integrations.list-tags
,list-tag-values
: Lists and describes tags available for cost filtering.list-anomalies
: Detects anomalies in cost reports.list-cost-providers
: Lists all cost providers available for filtering costs in VQL queries.
Getting Started: Vantage MCP Server Installation
Before diving into integrations, let's get the Vantage MCP Server up and running.
Prerequisites
- Go
- Node.js
- Read-Only Vantage API Token: Generate a new token specifically for the MCP Server.
Installation Methods
-
Using Homebrew: The easiest method for macOS users.
-
From Source: Offers more customization.
Debugging
Use the MCP inspector to test the installation:
Integrating with MCP Clients: Claude, Cursor, and Goose
Now for the exciting part: connecting the MCP Server to your favorite AI tools.
Claude for Desktop Setup
-
Download Claude for Desktop.
-
Open Settings (Cmd + , or Claude > Settings).
-
Go to Developer > Edit Config.
-
Edit
claude_desktop_config.json
: -
Restart Claude.
Cursor IDE Setup
-
Download Cursor.
-
Open Settings (Cursor > Settings > Cursor Settings).
-
Select MCP > Add new global MCP Server.
-
Edit
mcp.json
with the same configuration as above:
Goose AI Setup
-
Download Goose.
-
Open Settings (Goose > Settings).
-
Under Extensions, click Add custom extension.
-
Fill in the fields:
- ID:
vantage-mcp-server
- Name:
Vantage
- Description:
Query costs and usage data
- Command: Path to the Vantage MCP Server binary.
- Environment Variables: Add
VANTAGE_BEARER_TOKEN
with your Vantage API token.
- ID:
Now, you can ask Goose about your cloud costs using natural language.
Contributing
The Vantage team welcomes contributions.
- Fork the repository.
- Create a new branch.
- Make your changes.
- Submit a pull request.
Let's make cloud cost analysis more accessible together with the Vantage MCP Server!
Troubleshooting on MacOS
If you encounter an error message stating that the “developer cannot be verified”, resolve this by:
- Opening System Settings
- Navigating to “Privacy and Security”
- Scrolling to the bottom and clicking “Open Anyway” after you see the message that “vantage-mcp-server-macos” was blocked.
This should make the binary executable.