AWS Cost Explorer and Amazon Bedrock Model Invocation Logs MCP Server & Client

What is AWS Cost Explorer and Amazon Bedrock Model Invocation Logs MCP Server & Client?

This MCP server provides natural-language access to AWS Cost Explorer data and Amazon Bedrock model invocation logs stored in Amazon CloudWatch. It connects Claude Desktop or any MCP‑enabled client to AWS billing and usage APIs, enabling conversational analysis of spend and model usage.

How to use AWS Cost Explorer and Amazon Bedrock Model Invocation Logs MCP Server & Client?

Install Python 3.12 and uv, clone the repository, create a virtual environment, install dependencies, and configure AWS credentials. Run server.py with environment variables MCP_TRANSPORT (set to stdio for local or sse for remote), BEDROCK_LOG_GROUP_NAME, and optionally CROSS_ACCOUNT_ROLE_NAME. For local use, configure Claude Desktop via Docker or UV arguments. For remote use, deploy the server on EC2 and connect via SSE transport with a CLI client or Chainlit app.

Key features of AWS Cost Explorer and Amazon Bedrock Model Invocation Logs MCP Server & Client

• Amazon EC2 spend analysis for the previous day
• Amazon Bedrock usage broken down by region, user, and model
• Service‑level spend reports across all AWS services
• Granular cost breakdown by day, region, service, and instance type
• Cross‑account spend queries via IAM role assumption
• Both local (stdio) and remote (SSE over HTTPS) deployment modes

Use cases of AWS Cost Explorer and Amazon Bedrock Model Invocation Logs MCP Server & Client

• Ask “What was my EC2 spend yesterday?” and get a natural‑language answer
• Analyze Bedrock model invocation costs over the last 30 days per user and region
• Compare month‑over‑month cost increases for specific AWS services
• Retrieve detailed spend data from multiple AWS accounts using a single MCP server
• Build a LangGraph agent with MCP tools to automate cloud cost investigations

FAQ from AWS Cost Explorer and Amazon Bedrock Model Invocation Logs MCP Server & Client

What data sources does this server use?

It reads data from AWS Cost Explorer API and Amazon Bedrock model invocation logs stored in Amazon CloudWatch Logs.

What are the runtime requirements?

Python 3.12, uv, and AWS credentials with read‑only access to Cost Explorer and CloudWatch Logs. Optionally, Amazon Bedrock for the LangGraph agent and EC2 for remote hosting.

Can I query spend from other AWS accounts?

Yes. Set the CROSS_ACCOUNT_ROLE_NAME environment variable, then provide the target AWS account ID in your query. The server assumes the role in that account.

Does the remote MCP server support HTTPS?

Yes. See the “secure remote MCP server” section in the README for HTTPS setup. Note that MCP itself does not include authentication; do not transmit sensitive data over plain MCP.

What transport modes are supported?

Local mode uses stdio (stdin/stdout). Remote mode uses sse (server‑sent events) over TCP port 8000. Claude Desktop currently only supports stdio.