mirror of
https://github.com/awslabs/amazon-bedrock-agentcore-samples.git
synced 2025-09-08 20:50:46 +00:00
50 lines
3.2 KiB
Markdown
50 lines
3.2 KiB
Markdown
|
# Hosting MCP server on AgentCore Runtime
|
||
|
|
|
||
|
|
## Overview
|
||
|
|
|
||
|
|
In this session we will discuss how to host MCP tools on Amazon Bedrock AgentCore Runtime.
|
||
|
|
|
||
|
|
We will use the Amazon Bedrock AgentCore Python SDK to wrapper the agents function as an MCP server compatible with Amazon Bedrock AgentCore.
|
||
|
|
It will handle the MCP server details so you can focus on your agent's core functionality.
|
||
|
|
|
||
|
|
The Amazon Bedrock AgentCore Python SDK prepares your agent or tool code to run on AgentCore Runtime.
|
||
|
|
|
||
|
|
It will transform your code into the AgentCore standardized HTTP protocol or MCP protocol contracts to allow for direct REST API endpoint communication for a traditional request/response pattern (HTTP protocol) or Model Context Protocol for tools and agents servers (MCP Protocol).
|
||
|
|
|
||
|
|
When you are hosting tools, the Amazon Bedrock AgentCore Python SDK will implement the [Stateless Streamable HTTP] transport protocol with the `MCP-Session-Id` header for [session isolation]https://modelcontextprotocol.io/specification/2025-06-18/basic/transports#session-management, servers must support stateless operation to not reject platform generated Mcp-Session-Id header.
|
||
|
|
Your MCP server will then be hosted on port `8000` and will provide one invocation path: the `mcp-POST`. This interaction endpoint with receive the MCP RPC messages and process them through your tool's capabilities. It supports both application/json and text/event-stream as response content-types.
|
||
|
|
|
||
|
|
When you set your AgentCore protocol to MCP, AgentCore Runtime will expect the MCP server container to be on path `0.0.0.0:8000/mcp` as that's the default path supported by most of the official MCP server SDKs.
|
||
|
|
|
||
|
|
AgentCore Runtime requires you to host stateless streamable-http servers because it provides session-isolation by default and automatically adds a Mcp-Session-Id header for any request without it, so MCP clients can have continuity of connection to same Bedrock AgentCore Runtime session ID.
|
||
|
|
|
||
|
|
Payload of `InvokeAgentRuntime` API is completely pass through, so RPC messages of protocols like MCP can easily be proxied.
|
||
|
|
|
||
|
|
In this tutorial you will learn:
|
||
|
|
|
||
|
|
* How to create an MCP server with tools
|
||
|
|
* How to test your server locally
|
||
|
|
* How to deploy your server to AWS
|
||
|
|
* How to invoke your deployed server
|
||
|
|
|
||
|
|
### Tutorial Details
|
||
|
|
|
||
|
|
| Information | Details |
|
||
|
|
|:--------------------|:----------------------------------------------------------|
|
||
|
|
| Tutorial type | Hosting Tools |
|
||
|
|
| Tool type | MCP server |
|
||
|
|
| Tutorial components | Hosting tool on AgentCore Runtime. Creating an MCP server |
|
||
|
|
| Tutorial vertical | Cross-vertical |
|
||
|
|
| Example complexity | Easy |
|
||
|
|
| SDK used | Amazon BedrockAgentCore Python SDK and MCP Client |
|
||
|
|
|
||
|
|
### Tutorial Architecture
|
||
|
|
In this tutorial we will describe how to deploy an existing MCP server to AgentCore runtime.
|
||
|
|
|
||
|
|
For demonstration purposes, we will use a very simple MCP server with 3 tools: `add_numbers`, `multiply_numbers` and `greet_users`
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
### Tutorial Key Features
|
||
|
|
|
||
|
|
* Hosting MCP Server
|