aws-cost-explorer-mcp-server

aws-cost-explorer-mcp-server

106
research_and_datacloud_platforms

This project is an MCP server client system tailored to analyze AWS spending data and enable natural language querying through Anthropic's Claude. It supports local and remote operation, offering detailed insights into AWS spending patterns.

Overview

An MCP server and client designed for analyzing AWS spending data using AWS Cost Explorer and Amazon Bedrock model invocation logs. It interfaces with Anthropic's Claude for natural language queries. Key features include:

  • Amazon EC2 spend analysis for daily insights.
  • Amazon Bedrock spend analysis by region, users, and models for the last 30 days.
  • Comprehensive AWS service spending reports.
  • Interactive, natural language query interface through Claude.

Features

  • EC2 spending breakdowns.
  • Regional and service-based spending analysis.
  • Detailed cost data by day, region, service, and instance type.
  • Claude Desktop compatibility for local and remote server setups.

Requirements

  • Python 3.12, AWS credentials, Anthropic API access, and optionally Amazon EC2 and Amazon Bedrock.

Installation

Install uv, clone the repository, set up a Python virtual environment, and configure AWS credentials.

Usage

Set up model invocation logs in Amazon CloudWatch, ensure IAM user/role has necessary access, configure for local or remote use, and start the server.

Development

Includes Docker support for containerized deployment, and guidelines for extending functionality with new cost analysis tools.