ProxyMCP Configuration System

This document describes how to use the ProxyMCP.from_config() factory method to create ProxyMCP instances from configuration files.

Overview

The configuration system uses a clean, modular architecture where each object type is responsible for creating itself from configuration. The factory method supports creating instances with any type currently supported in src/mcp_kit/:

Architecture

This design follows the single responsibility principle, where each target is responsible for understanding its own configuration format.

Configuration Format

Configuration files support both standard YAML or JSON formats.

Using Target Factory Methods Directly

You can also create targets directly using their individual factory methods:

from omegaconf import OmegaConf
from src.mcp_kit.factory import create_target_from_config

# Create targets directly
config = OmegaConf.load("config.yaml")

# Direct target creation
target = create_target_from_config(config.target)

# Then wrap in ProxyMCP
proxy = ProxyMCP(mcp_target)

Example Configuration Files

Example configuration files are available in examples/proxy_configs/:

• mcp_target.yaml - Simple MCP target
• oas_target.yaml - OpenAPI specification target
• mocked_random_target.yaml - Mocked target with random responses
• mocked_llm_target.yaml - Mocked target with LLM responses
• multiplex_target.yaml - Multiplex target combining multiple servers
• mcp_target.json - JSON format example

Error Handling

The factory method will raise ValueError for:

• Unknown target types
• Unknown response generator types
• Missing required configuration fields

Testing

Unit tests are available for both the high-level ProxyMCP factory and individual target factories:

ProxyMCP factory tests:

uv run pytest tests/test_proxy_config.py::TestProxyMCPFromConfig -v

Individual target factory tests:

uv run pytest tests/test_target_from_config.py::TestTargetFromConfig -v

Run all configuration tests:

uv run pytest tests/test_proxy_config.py tests/test_target_from_config.py -v

The test suite covers:

• All target types and their factory methods
• Error handling for invalid configurations
• Nested target configurations (e.g., mocked targets with mocked base targets)
• Both YAML and JSON configuration formats
• Edge cases like minimal configurations and default values

Future Enhancements

The configuration system is designed to be easily extensible for future features like:

• Multiple proxy instances from a single configuration file
• Environment variable substitution
• Configuration validation schemas
• Dynamic configuration reloading