Lambda MCP Server Demo (Streamable HTTP)

What is Lambda MCP Server Demo (Streamable HTTP)?

This project demonstrates a developer-friendly way to create serverless MCP (Model Context Protocol) tools using AWS Lambda. It showcases a stateless, serverless MCP server with minimal boilerplate, and includes a TypeScript client that integrates with Amazon Bedrock Converse API using Amazon Nova Pro. It is designed for developers who want to deploy cloud-hosted MCP tools via Streamable HTTP.

How to use Lambda MCP Server Demo (Streamable HTTP)?

Install prerequisites (AWS account, AWS SAM CLI, Node.js, Docker, Python 3.9+, Amazon Bedrock access). Deploy the server with sam build then sam deploy --guided, providing an McpAuthToken. Navigate to the client directory and run run-client.sh. Add new tools using the @mcp_server.tool() decorator in server/app.py.

Key features of Lambda MCP Server Demo (Streamable HTTP)

• Uses Streamable HTTP transport (not SSE).
• Minimal boilerplate with decorator-based tool creation.
• Built-in session state management via DynamoDB.
• Bearer token authentication compliant with MCP standard.
• Example tools: get_time, get_weather, count_s3_buckets.
• Client integration with Amazon Bedrock Converse API.

Use cases of Lambda MCP Server Demo (Streamable HTTP)

• Deploying serverless MCP agents with AWS Lambda.
• Building stateless, session-aware MCP tools for cloud environments.
• Demonstrating Streamable HTTP MCP communication with Bedrock.
• Rapid prototyping of MCP tools with minimal setup and infrastructure.

FAQ from Lambda MCP Server Demo (Streamable HTTP)

What MCP clients support Streamable HTTP?

Very few MCP clients currently support Streamable HTTP. The included TypeScript client uses the TypeScript MCP SDK, which does support Streamable HTTP. For other clients, check compatibility.

What are the runtime dependencies and requirements?

Requires an AWS account, AWS SAM CLI, Node.js and npm, Docker, Python 3.9+, and access to Amazon Bedrock with Amazon Nova Pro enabled. The server is deployed via SAM; the client runs in Docker.

Where does session data live?

Session state is stored in a DynamoDB table against a sessionId key. This persistence is automatically managed across tool invocations within the same conversation.

Is this production ready?

No, this is a proof-of-concept implementation. The bearer token authentication is primarily for demonstration and development. For production, consider adding IAM roles, OAuth 2.0, or Amazon Cognito.

How does authentication work?

The server uses Bearer token authentication via an Authorization header. A custom authorizer in API Gateway validates the token. The client automatically includes the token when provided. The token is stored locally in .mcp-api-token (for convenience only, not production secure).