Provides MCP servers that connect AWS docs, tooling, and data sources to AI workflows.
Configuration
View docs{
"mcpServers": {
"knowledge_mcp": {
"url": "https://knowledge-mcp.global.api.aws"
}
}
}You use MCP (Model Context Protocol) servers to connect AWS documentation, data sources, and AWS-specific tooling to your AI workflows. These servers expose capabilities that a client app can access to enhance coding, cloud operations, and integration tasks with up-to-date AWS content and specialized tooling.
Choose an MCP client you trust (for example, a coding assistant that supports MCP). Add the MCP servers you need to your client configuration, using either the remote HTTP servers or local stdio servers described here. When you issue a task, the client will connect to the configured MCP servers to fetch AWS documentation, guidance, or tooling context as part of your AI workflow. You can reference AWS services, run infrastructure-as-code tasks, or retrieve product documentation directly through your AI assistant.
Typical usage patterns include: selecting a knowledge MCP to retrieve official AWS docs while you code, enabling an IaC MCP to guide CDK/Terraform tasks, or using a data/analytics MCP to assist with database operations. Ensure your client has network access to any HTTP servers and that local MCP servers are running on your machine if you opt for stdio configurations.
Prerequisites: install the required runtime tools and ensure your environment has network access to the MCP endpoints.
1) Install uvx or equivalent MCP client helpers and Python or Node as required by the local servers you plan to run.
2) Add MCP server entries to your client configuration. If you are using an MCP client that reads a JSON configuration, you will include the following entries (examples shown below).
3) Start any stdio MCP servers you plan to run locally, following the exact commands shown in the examples. If a server is remote, you only need the HTTP URL and do not start a local process.
This guide includes a couple of concrete MCP server configurations that are explicitly provided as part of the setup examples.
{
"mcpServers": {
"core_mcp": {
"type": "stdio",
"command": "uvx",
"args": ["awslabs.core-mcp-server@latest"],
"env": {
"FASTMCP_LOG_LEVEL": "ERROR"
}
},
"core_mcp_win": {
"type": "stdio",
"command": "uv",
"args": ["tool","run","--from","awslabs.core-mcp-server@latest","awslabs.core-mcp-server.exe"],
"env": {
"FASTMCP_LOG_LEVEL": "ERROR"
}
},
"knowledge_mcp": {
"type": "http",
"name": "knowledge_mcp",
"url": "https://knowledge-mcp.global.api.aws",
"args": []
}
}
}