How does the Lambda function manage the MCP server lifecycle?
Each invocation starts the MCP server as a child process, initializes it, forwards the incoming request, returns the server’s response, and then shuts down the child process.
@awslabs
Basic information
Run existing Model Context Protocol (MCP) stdio-based servers in AWS Lambda functions
Add this server to your MCP-compatible client using the configuration below.
{
"mcpServers": {
"run-model-context-protocol-servers-with-aws-lambda": {
"command": "python",
"args": [
"-m",
"my_mcp_server_python_module",
"--my-server-command-line-parameter",
"some_value"
]
}
}
}Tool list not available yet
This listing does not currently have a tool list to display. This does not mean the server has no tools; check its documentation for supported capabilities.
We auto-extract tools from the README. The maintainer can list them under a ## Tools heading to populate this section.
It enables you to run MCP stdio-based servers in AWS Lambda functions, wrapping existing MCP servers so that clients communicate over short-lived HTTPS connections instead of long-running local stdio streams. This library targets developers who want to deploy MCP servers in a distributed cloud architecture while still using standard MCP tooling.
Install the package from PyPI (run-mcp-servers-with-aws-lambda) or NPM (@aws/run-mcp-servers-with-aws-lambda). Determine your MCP server’s command and arguments (using StdioServerParameters in Python or serverParams in TypeScript), then package the server with the Lambda function code. Use the provided handler classes (e.g., APIGatewayProxyEventHandler) to process incoming MCP requests. The README includes examples for both Python and TypeScript, as well as guidance on passing secrets and AWS credentials to the child process.
Each invocation starts the MCP server as a child process, initializes it, forwards the incoming request, returns the server’s response, and then shuts down the child process.
The library supports OAuth via API Gateway (e.g., Amazon Cognito), SigV4 authentication with Lambda function URLs, and IAM authentication when using the Lambda Invoke API.
You can configure secrets as encrypted Lambda environment variables or retrieve them from AWS Secrets Manager in the function code. The README provides code examples for both Python and TypeScript. Note that any user with invoke access can also use those secrets.
Four transports are supported: MCP Streamable HTTP via API Gateway, MCP Streamable HTTP via Amazon Bedrock AgentCore Gateway, a custom Streamable HTTP with SigV4 for Lambda function URLs, and a custom Lambda invocation transport using

Web hosting for websites built with AI. Connect your HostingFor.AI account and your AI assistant can publish and maintain your site directly: upload files, manage DNS records for custom domains, run SQL against your MySQ

InfraNest watches your websites, servers, domains and SSL certificates. Connect it to Claude, ChatGPT or Cursor and ask what is down or about to expire.
YesTechie builds enterprise UniFi networks, cameras, access control, cabling and Wi-Fi for businesses nationwide from its California HQ. Agents can read real project case studies and practical configuration guides, price
MCP server connecting to Kubernetes
MCP server for Docker
Comments