# MCP Server POC for Amazon Bedrock AgentCore
This POC demonstrates deployment and management of Model Context Protocol (MCP) servers on Amazon Bedrock AgentCore using the Gateway method.
## Overview
The Model Context Protocol (MCP) provides a standardized interface for AI applications to access tools, resources, and enterprise data sources. This solution implements:
- **Gateway Method**: Wraps existing Lambda functions or REST APIs as MCP tools, handling protocol translation automatically
- **MCP Server**: Direct implementation using FastMCP framework
## Architecture
```
┌─────────────────┐
│ Bedrock │
│ AgentCore │
└────────┬────────┘
│
▼
┌─────────────────┐
│ MCP Gateway │ ← Protocol Translation
│ (Lambda) │
└────────┬────────┘
│
┌────┴────┐
▼ ▼
┌────────┐ ┌────────┐
│Lambda │ │ REST │
│Functions│ │ APIs │
└────────┘ └────────┘
```
## Components
1. **MCP Gateway** (`gateway/`): Translates MCP protocol to Lambda/REST API calls
2. **MCP Server** (`server/`): Direct MCP server implementation using FastMCP
3. **Example Tools** (`tools/`): Sample Lambda functions and REST APIs
4. **Infrastructure** (`infrastructure/`): AWS SAM templates for deployment
## Prerequisites
- Python 3.11+
- AWS CLI configured
- AWS SAM CLI installed
- Docker (for local testing)
## Quick Start
### 1. Install Dependencies
```bash
pip install -r requirements.txt
```
### 2. Deploy Infrastructure
**Option A: Using AWS SAM (Recommended for beginners)**
```bash
cd infrastructure
sam build
sam deploy --guided
```
**Option B: Using Terraform**
```bash
cd terraform
cp terraform.tfvars.example terraform.tfvars
# Edit terraform.tfvars with your values
terraform init
terraform plan
terraform apply
```
See `TERRAFORM_QUICKSTART.md` for detailed Terraform deployment guide with Python dependencies layer configuration.
### 3. Test MCP Gateway
```bash
python test_gateway.py
```
## Project Structure
```
mcp_1/
├── gateway/ # MCP Gateway implementation
│ ├── __init__.py
│ ├── gateway.py # Main gateway handler
│ └── protocol.py # MCP protocol translation
├── server/ # Direct MCP server
│ ├── __init__.py
│ └── mcp_server.py # FastMCP server
├── tools/ # Example tools
│ ├── lambda_functions/ # Lambda functions as MCP tools
│ └── rest_apis/ # REST API endpoints
├── infrastructure/ # AWS SAM deployment
│ ├── template.yaml # SAM template
│ └── README.md # SAM deployment guide
├── terraform/ # Terraform deployment
│ ├── main.tf # Main Terraform configuration
│ ├── variables.tf # Variable definitions
│ ├── outputs.tf # Output values
│ ├── versions.tf # Provider versions
│ ├── README.md # Terraform deployment guide
│ └── Makefile # Helper commands
├── TERRAFORM_QUICKSTART.md # Quick Terraform deployment guide
├── QUICKSTART.md # Local development quick start
├── tests/ # Test files
├── requirements.txt # Python dependencies
└── README.md # This file
```
## Gateway Method
The Gateway method automatically:
- Translates MCP tool calls to Lambda invocations
- Translates MCP tool calls to REST API requests
- Handles authentication and error handling
- Provides unified interface for multiple backends
## Example Usage
### Registering a Lambda Function as MCP Tool
```python
from gateway.gateway import MCPGateway
gateway = MCPGateway()
gateway.register_lambda_tool(
name="weather_lookup",
lambda_arn="arn:aws:lambda:us-east-1:123456789:function:WeatherFunction",
description="Get weather information for a location"
)
```
### Registering a REST API as MCP Tool
```python
gateway.register_rest_tool(
name="stock_price",
endpoint="https://api.example.com/stock",
method="GET",
description="Get stock price information"
)
```
## Testing
Run the test suite:
```bash
pytest tests/
```
Test the gateway locally:
```bash
python -m gateway.gateway
```
## Deployment
- **Terraform**: See `TERRAFORM_QUICKSTART.md` for quick Terraform deployment (includes Python dependencies layer)
- **SAM**: See `DEPLOYMENT.md` for detailed SAM deployment instructions
- **Terraform Details**: See `terraform/README.md` for comprehensive Terraform guide
## License
MIT
