mcpmesh/python-runtime repository overview

MCP Mesh Python Runtime

Base container image for building distributed Model Context Protocol (MCP) agents with Python

The MCP Mesh Python Runtime provides a pre-configured environment for developing and deploying MCP agents that automatically integrate with the MCP Mesh service discovery system.

🚀 Quick Start

⁠Use as base image in your Dockerfile

FROM mcpmesh/python-runtime:latest

⁠Copy your agent code

COPY my_agent.py .

⁠Your agent automatically connects to MCP Mesh registry

CMD ["python", "my_agent.py"]

🏗️ What's Included

• Python 3.11 - Modern Python runtime
• MCP Mesh SDK - Full service mesh integration
• FastMCP - High-performance MCP protocol implementation
• Auto-Discovery - Automatic registry registration and capability injection
• Health Monitoring - Built-in health checks and graceful shutdown
• Dependency Injection - Dynamic remote function calls via HTTP proxies

🛠️ Agent Development

Create agents using simple decorators:

import mesh

@mesh.agent( name="my-agent", capabilities=["data_processing", "file_operations"], dependencies=["database", "storage"] ) class MyAgent: @mesh.tool def process_data(self, data: str) -> dict: # Your agent logic here return {"processed": data}

📦 Environment Variables

🔄 Runtime Features

• Resilient Connection: Works offline, auto-connects when registry available
• Hot Swapping: Dynamic capability updates without restarts
• Graceful Degradation: Continues operation if registry fails
• Microservices Ready: Designed for Kubernetes and Docker Compose

📋 Supported Architectures

• linux/amd64 - Intel/AMD 64-bit
• linux/arm64 - ARM 64-bit (Apple Silicon, Graviton)

🔗 Links

• Documentation: https://github.com/dhyansraj/mcp-mesh/tree/main/docs⁠
• Examples: https://github.com/dhyansraj/mcp-mesh/tree/main/examples⁠
• Source Code: https://github.com/dhyansraj/mcp-mesh⁠
• PyPI Package: https://pypi.org/project/mcp-mesh/⁠

Part of the MCP Mesh ecosystem - see also https://hub.docker.com/r/mcpmesh/registry⁠ and https://hub.docker.com/r/mcpmesh/cli⁠.

---

This focuses on the developer experience and shows how easy it is to build agents using the runtime image.