AWS Cost Explorer MCP Server

Overview

The Sample Cloud Spend MCP Server is a specialized bridge that lets Claude and other AI assistants interrogate AWS billing data in real time. By exposing the Cost Explorer API and CloudWatch Logs through the Model Context Protocol, it turns raw spend numbers into conversational insights. Developers can ask natural‑language questions such as “How much did we spend on EC2 in the last 24 hours?” or “Show me Bedrock usage by region for this month.” and receive structured, up‑to‑date answers without writing SQL or manually exporting reports.

At its core, the server translates MCP queries into authenticated AWS API calls. It retrieves cost and usage metrics, aggregates them by service, region, or instance type, and returns the results in a format Claude can ingest. Because it runs over HTTPS, it can be hosted on an EC2 instance or any other secure environment, allowing teams to keep sensitive billing data in‑house while still benefiting from AI-driven analysis. The integration with Bedrock model invocation logs further enriches the dataset, enabling cost attribution to specific models and users—an invaluable capability for organizations experimenting with generative AI workloads.

Key capabilities include:

• Granular spend breakdowns for EC2, Bedrock, and all other AWS services over configurable windows (e.g., last day or month).
• Region‑level and user‑level visibility for Bedrock, helping teams understand where and how AI models are being consumed.
• Interactive querying via Claude Desktop or any MCP‑compatible client, eliminating the need for manual report generation.
• Secure remote operation, with optional HTTPS support and IAM‑based access controls that align with AWS best practices.

Typical use cases span financial operations, cost optimization teams, and AI product managers. A finance analyst might ask for a quick audit of last month’s spend to feed into budgeting models, while an AI platform owner could track Bedrock usage per user to enforce quotas or detect anomalous activity. In both scenarios, the MCP server provides a frictionless, conversational interface that turns complex billing data into actionable insights.

By integrating seamlessly with existing AI workflows—whether through Claude Desktop or a LangGraph Agent—the Sample Cloud Spend MCP Server empowers developers to embed cost awareness directly into their conversational agents. This not only streamlines decision‑making but also promotes a culture of cost transparency across engineering and product teams.