by vantage-sh
Fetch cloud cost and usage data from Vantage via natural‑language queries, enabling AI assistants and MCP clients to retrieve, filter, and analyze spend information.
The server acts as a bridge between Vantage's cost APIs and Model Context Protocol (MCP) clients. It translates natural‑language requests from tools such as Claude, Cursor, or Goose into VQL queries, returning structured cost, budgeting, and forecasting data.
.tool-versions
) and Node.js. Generate a read‑only Vantage API token.go build -o vantage-mcp-server
, and make the binary executable.npx @modelcontextprotocol/inspector -e VANTAGE_BEARER_TOKEN=<token> ./vantage-mcp-server
).VANTAGE_BEARER_TOKEN
environment variable as shown in the README.Q: Do I need a write‑enabled token? A: No. A read‑only Vantage API token is sufficient; write operations are not supported.
Q: Can I run this server remotely? A: This repository provides the self‑hosted version. For a remote service, refer to the official Vantage MCP documentation.
Q: Which platforms are supported? A: Any OS that can run a Go binary. macOS users may need to allow the executable in System Settings if macOS blocks it.
Q: How do I update the server after pulling new changes?
A: Re‑run go build
to compile the latest version.
Q: What transport does the server use? A: Standard Input/Output (stdio) transport, compatible with MCP clients that support it.
The Vantage MCP Server is an open-source tool, written in Golang, that lets you interact with your cloud cost data through AI assistants and MCP clients. By acting as a bridge to Vantage's existing APIs, the Vantage MCP Server lets you query cloud spend data using natural language and makes cost analysis more intuitive.
📝 Note: The Vantage MCP Server is available in both self-hosted and remote forms. This repository supports the self-hosted version, which runs locally using Standard Input/Output (stdio) Transport. For a remote alternative, see the Vantage MCP documentation.
The Vantage MCP Server currently exposes the following tools, which can be invoked by any compatible MCP client (e.g., Claude, Cursor, Goose):
query-costs
list-costs
list-cost-reports
get-cost-report-forecast
list-cost-integrations
list-cost-providers
list-cost-services
list-budgets
list-dashboards
list-tags
list-tag-values
list-anomalies
list-unit-costs
get-myself
submit-user-feedback
If you're installing from source, ensure the following packages are installed (see .tool-versions
for exact versions):
You can use a version manager (e.g., asdf
) or package manager (e.g., Homebrew) to install these dependencies.
You will also need to create a Read-Only Vantage API token (Write will not work at this time). Follow the instructions on the Vantage API documentation. We recommend creating a brand-new read-only API token for exclusive use with the MCP Server.
brew install vantage-sh/tap/vantage-mcp-server
git clone https://github.com/vantage-sh/vantage-mcp-server
go build -o vantage-mcp-server
chmod +x vantage-mcp-server
📝 Note: If you pull down new changes from the repository, be sure to re-run
go build
to rebuild the server and ensure you're running the latest version.
npx @modelcontextprotocol/inspector -e VANTAGE_BEARER_TOKEN=<token> ./vantage-mcp-server
Setup instructions vary depending on which MCP client you use. Example clients include:
See the MCP documentation for a list of available clients. Detailed instructions for Claude for Desktop, Cursor, and Goose are provided below.
Command + ,
).~/Library/Application Support/Claude/claude_desktop_config.json
%APPDATA%\Claude\claude_desktop_config.json
Open the claude_desktop_config.json
file and update its contents. Make sure to replace the placeholders <path_to_compiled_vantage_mcp_server_binary>
with the path where you downloaded the Vantage MCP Server binary, and <personal_vantage_api_token>
with your Vantage API token.
{
"mcpServers": {
"Vantage": {
"command": "<path_to_compiled_vantage_mcp_server_binary>",
"args": [],
"env": { "VANTAGE_BEARER_TOKEN": "<personal_vantage_api_token>" }
}
}
}
Save the configuration file and restart Claude.
In the bottom-right corner of the Claude for Desktop input box, click the hammer icon to see the available tools for the Vantage MCP Server.
Once you've set up the configuration, you can start prompting Claude. Each time you use a new tool, Claude will ask for your approval before proceeding.
Download Cursor.
Open Cursor and click Cursor > Settings > Cursor Settings from the menu bar.
In the left pane, select MCP.
Click Add new global MCP Server.
Update the contents of the opened mcp.json
file. Make sure to replace the placeholders <path_to_compiled_vantage_mcp_server_binary>
with the path where you downloaded the Vantage MCP Server binary, and <personal_vantage_api_token>
with your Vantage API token.
{
"mcpServers": {
"Vantage": {
"command": "<path_to_compiled_vantage_mcp_server_binary>",
"args": [],
"env": { "VANTAGE_BEARER_TOKEN": "<personal_vantage_api_token>" }
}
}
}
Command + ,
).vantage-mcp-server
.Vantage
.Query costs and usage data
.VANTAGE_BEARER_TOKEN
and the value set to your Vantage API token.If you download a release from our Github page and the executable fails to run "because the developer cannot be verified", please open your System Settings. Then find the "Privacy and Security" section. Then scroll to the bottom and you should see a message that "vantage-mcp-server-macos" was blocked, click the "open anyway" button. After this flow, the executable should be able to be run without issue.
If you'd like to contribute to this project:
git checkout -b feature/my-feature
.We welcome community contributions, improvements, and bug fixes. If you run into any issues, submit a bug report via this repository's GitHub Issues.
See the LICENSE.MD
file for commercial and non-commercial licensing details.
Please log in to share your review and rating for this MCP.
Discover more MCP servers with similar functionality and use cases
by daytonaio
Provides a secure, elastic sandbox environment for executing AI‑generated code with isolated runtimes and sub‑90 ms provisioning.
by awslabs
Specialized servers that expose AWS capabilities through the Model Context Protocol, allowing AI assistants and other applications to retrieve up‑to‑date AWS documentation, manage infrastructure, query services, and perform workflow automation directly from their context.
by awslabs
AWS MCP Servers allow AI agents to interact with and manage a wide range of AWS services using natural language commands. They enable AI-powered cloud management, automated DevOps, and data-driven insights within the AWS ecosystem.
by cloudflare
Remote Model Context Protocol endpoints that let AI clients read, process, and act on data across Cloudflare services such as Workers, Radar, Observability, and more.
by supabase-community
Enables AI assistants to interact directly with Supabase projects, allowing them to query databases, fetch configuration, manage tables, and perform other project‑level operations.
by Azure
azure-mcp is a server that implements the Model Context Protocol (MCP) to connect AI agents with Azure services. It allows developers to interact with Azure resources like Storage, Cosmos DB, and the Azure CLI using natural language commands within their development environment.
by Flux159
MCP Server for Kubernetes management commands, enabling interaction with Kubernetes clusters to manage pods, deployments, and services.
by strowk
Provides a Golang‑based server that enables interaction with Kubernetes clusters via prompts, allowing listing of contexts, namespaces, resources, nodes, pods, events, logs, and executing commands inside pods.
by jamsocket
Run arbitrary Python code securely in persistent, stateful sandboxes that remain available indefinitely.