Feature/llm gateway

[darielnoel]

Secure API Key Management and LLM Proxy Integration

Description

This PR addresses the critical issue of secure API key management in frontend applications, as outlined in Issue #78. It implements a solution that allows developers to securely handle API keys for various LLM providers without exposing them in client-side code.

Changes

1. Enhanced BaseAgent class with improved normalizeLlmConfig method:

   • Standardizes configuration across different LLM providers
   • Supports custom apiBaseUrl for each provider
   • Maintains flexibility for rapid prototyping and production environments

2. Integration with Kaiban LLM Proxy:

   • Demonstrates secure routing of API requests
   • Provides example implementation for OpenAI, Anthropic, Google Gemini, and Mistral

Usage Example

const agent = new Agent({
   name: 'AI Assistant',
   role: 'Helper',
   llmConfig: {
      provider: "openai",
      apiBaseUrl: "http://localhost:3000/llm/openai",
      model: "gpt-4",
   }
});

Benefits

• Enhances security by keeping API keys server-side
• Simplifies development process, especially for prototyping
• Provides a flexible solution adaptable to various environments
• Maintains compatibility with multiple LLM providers

Testing

• Tested with local setup of Kaiban LLM Proxy
• Verified functionality with OpenAI, Anthropic, Google, and Mistral providers

Documentation

• Updated README with usage instructions and examples
• Added comments in code for clarity on configuration options

Related

• Closes #78
• References Kaiban LLM Proxy