LLM Configuration Examples

This directory contains example configurations for different LLM providers.

Quick Start

1. Choose your provider:

   • openai_config.json - OpenAI GPT models (requires API key)
   • anthropic_config.json - Anthropic Claude models (requires API key)
   • ollama_config.json - Ollama local models (free, requires Ollama installed)

2. Copy to your config directory:

   # For macOS/Linux
   cp openai_config.json ~/.mcpproxy/mcp_config.json
   
   # Or for Windows
   copy openai_config.json %USERPROFILE%\.mcpproxy\mcp_config.json

3. Add your API key (OpenAI or Anthropic):

   • Edit the config file
   • Replace YOUR-KEY-HERE with your actual API key

4. Restart mcpproxy:

   pkill mcpproxy
   ./mcpproxy serve

Provider Details

OpenAI (openai_config.json)

• Pros: Fast, reliable, good general performance
• Cons: Requires paid API key
• Best for: Most users, production use
• Get API key: https://platform.openai.com/api-keys

Anthropic (anthropic_config.json)

• Pros: Excellent reasoning, advanced capabilities
• Cons: Requires paid API key
• Best for: Complex diagnostic tasks, advanced reasoning
• Get API key: https://console.anthropic.com/

Ollama (ollama_config.json)

• Pros: Free, fully local, privacy-focused
• Cons: Requires local installation, slower on weak hardware
• Best for: Privacy-sensitive environments, offline use
• Setup:

  # Install Ollama
  brew install ollama  # macOS
  # Or download from https://ollama.ai
  
  # Start Ollama
  ollama serve
  
  # Pull a model
  ollama pull llama2

Customization

All config files support these LLM parameters:

{
  "llm": {
    "provider": "openai|anthropic|ollama",
    "model": "model-name",
    "temperature": 0.7,
    "max_tokens": 2000
  }
}

Temperature

• 0.0: Deterministic, focused responses
• 0.7: Balanced (recommended)
• 1.0: Creative, varied responses

Max Tokens

• Controls response length
• Higher = longer responses (but more expensive)
• Default: 2000 tokens (~1500 words)

See Also

• Complete LLM Configuration Guide
• Main mcpproxy README