medhaai 0.1.0

A flexible multi-agent AI framework

MedhaAI

MedhaAI is an advanced AI agent framework with minimal abstraction, designed for easy integration of language models and web scraping capabilities.

Features

• Support for multiple LLM providers (OpenAI and Anthropic)
• Advanced web scraping tool
• Flexible agent system for complex task execution
• Built-in error handling and logging
• Easy configuration management

Installation

pip install medhaai

Quick Start

import asyncio
from medhaai import MedhaConfig, MedhaLLM, MedhaAgent, ToolKit

async def main():
    config = MedhaConfig(
        openai_api_key="your_openai_api_key_here",
        model_name="gpt-3.5-turbo"
    )
    llm = MedhaLLM(config)
    toolkit = ToolKit()
    agent = MedhaAgent(llm, toolkit)
    
    result = await agent.run("Research the latest advancements in quantum computing")
    print(result)

asyncio.run(main())

Documentation

For full documentation, visit docs.medhaai.com.

Advanced Usage

Using Different LLM Providers

MedhaAI supports both OpenAI and Anthropic models. Here's how to use an Anthropic model:

config = MedhaConfig(
    anthropic_api_key="your_anthropic_api_key_here",
    model_name="claude-2"
)
llm = MedhaLLM(config)

Custom Web Scraping

You can use the web scraping tool directly:

from medhaai import AdvancedWebScraperTool

scraper = AdvancedWebScraperTool()
result = await scraper.scrape("https://example.com")
print(result)

Error Handling

MedhaAI provides custom exceptions for better error handling:

from medhaai.exceptions import MedhaAIException

try:
    result = await agent.run("Some task")
except MedhaAIException as e:
    print(f"An error occurred: {str(e)}")

Contributing

Contributions are welcome! Please see our Contributing Guide for more details.

License

This project is licensed under the MIT License - see the LICENSE file for details.