Provides AWS knowledge, docs, and tooling through MCP to assist AI-enabled development workflows.
Configuration
View docs{
"mcpServers": {
"aws_knowledge_mcp": {
"url": "https://knowledge-mcp.global.api.aws"
}
}
}You use MCP servers to give your AI applications direct access to AWS data, docs, and tooling. This lets you query up-to-date AWS information, run AWS tasks, and embed AWS workflows into your AI-powered tools while keeping control of security, access, and workflows. By pairing a local or remote MCP server with your MCP client, you can provide context-rich capabilities to agents, IDEs, chatbots, and automation scripts that neatly extend your development and operational workflows.
Connect your MCP client to one or more MCP servers to extend your AI tools with AWS capabilities. You can use a remote, hosted MCP server for AWS knowledge and documentation, or run a local MCP server to expose a toolset from your own environment. Your client maintains 1:1 connections to MCP servers, enabling these servers to supply AWS context, guidance, and tooling to your AI workflows. Common use cases include real-time AWS documentation access during coding, infrastructure-as-code guidance, cost estimation prompts, and AWS service guidance embedded directly into your AI assistant.
Prerequisites: you need a working MCP client setup and the ability to run MCP servers locally or in your environment. Install the MCP client runtime and the required runtime tools for local servers (for example, a universal runner like uv). The instructions below describe a practical, supported path to get started with a simple local MCP server and a remote AWS knowledge server.
1) Install the MCP runtime and required tooling on your machine.
2) Add a local MCP server configuration using a standardized MCP config snippet for a core MCP server. This creates a stdio-based MCP server you can run locally.
3) Optionally add a remote AWS knowledge MCP server using an HTTP transport configuration to access up-to-date AWS docs and references.
Below are two example MCP server configurations you can use as a starting point. The first is a local stdio MCP server that runs via uvx. The second is a remote HTTP MCP server that exposes knowledge content from AWS documentation sources.
Control access to MCP servers through your chosen authentication method and apply least-privilege policies for AWS interactions. For local development, keep sensitive credentials and tokens on a secure machine, and prefer temporary credentials or profiles when possible. Review and adjust log levels and access controls via environment variables and server configuration.
If you encounter startup time or dependency issues, verify that all required runtimes and dependencies are installed and that your MCP client is configured to the correct tool name and version. Check your environment variables and ensure correct paths to your MCP settings files are provided.
General AWS API access and command validation for AWS service interactions through MCP servers
Infrastructure-as-code guidance and tooling to help you build and deploy AWS CDK constructs via MCP
Direct access to the latest AWS documentation and API references within your AI workflow
Cost estimates and pricing guidance for AWS services within MCP-enabled prompts
Bedrock knowledge bases retrieval and citation-enabled queries for Bedrock-based workflows