ReActLM 🤖

A powerful reactive agent framework powered by Language Models (LLMs). ReActLM provides a flexible architecture for building AI agents that can reason, use tools, and maintain memory of their actions.

🌟 Features

• 🧠 Advanced Reasoning - Chain of thought prompting with GPT-4 and other LLMs
• 🔍 Research Tools - Integrated Brave Search and Wikipedia research capabilities
• 💾 Persistent Memory - Redis-based memory system for maintaining context
• 📊 Structured Output - JSON-formatted responses for easy integration
• 🔄 Async First - Built with asyncio for high-performance operations
• 🛠️ Extensible - Plugin architecture for easy tool integration
• 📝 Detailed Tracing - Complete execution traces for debugging and analysis

🚀 Getting Started

Currently, this project is in development. To use it:

1. Clone the repository

git clone https://github.com/Prashant18/ReActLM.git
cd ReActLM

2. Set up your environment and build the project locally

Basic Usage

import asyncio
from reactlm import Agent, AgentConfig, OpenAILLM, BraveSearchTool

async def main():
    # Initialize agent
    agent = Agent(
        llm=OpenAILLM(api_key="your_openai_key"),
        config=AgentConfig(
            max_iterations=5,
            temperature=0.7
        )
    )
    
    # Add research capabilities
    await agent.add_tool(BraveSearchTool(api_key="your_brave_key"))
    
    # Execute a task
    result = await agent.execute(
        "What are the latest developments in quantum computing?"
    )
    print(result.content)

if __name__ == "__main__":
    asyncio.run(main())

🔑 API Keys Setup

ReActLM requires API keys for various services. Create a .env file in your project root:

# Required
OPENAI_API_KEY=your_openai_api_key_here
BRAVE_API_KEY=your_brave_api_key_here

# Optional (for memory persistence)
REDIS_HOST=localhost
REDIS_PORT=6379
REDIS_PASSWORD=your_redis_password  # If using authentication

⚠️ Security Note: Never commit your .env file to version control. Add it to your .gitignore:

# .gitignore
.env
*.pyc
__pycache__/

Obtaining API Keys

1. OpenAI API Key:

   • Sign up at OpenAI Platform
   • Navigate to API Keys section
   • Create a new secret key

2. Brave Search API Key:

   • Visit Brave Search API
   • Sign up for API access
   • Get your API key (2,000 free searches/month)

3. Redis Setup (Optional):

   • Local: Install Redis using package manager
   • Cloud: Use services like Redis Labs
   • Configure authentication if needed

🛠️ Components

LLM Providers

• OpenAI GPT-4/3.5
• Support for custom LLM implementations

Research Tools

• Brave Search Integration
• Wikipedia Research
• Extensible tool system

Memory Systems

• Redis-based persistence
• In-memory fallback
• Customizable storage backends

📚 Advanced Usage

Chain of Thought Research

from reactlm import Agent, AgentConfig, ExecutionMode

# Initialize agent with research configuration
agent = Agent(
    llm=OpenAILLM(
        api_key=os.getenv("OPENAI_API_KEY"),
        model="gpt-4"
    ),
    config=AgentConfig(
        max_iterations=5,
        mode=ExecutionMode.STANDARD,
        metadata={"purpose": "research"}
    )
)

# Add research tools
await agent.add_tool(BraveSearchTool())
await agent.add_tool(WikipediaTool())

# Execute research with context
result = await agent.execute(
    "Analyze quantum computing developments",
    context={
        "format": "structured",
        "depth": "comprehensive",
        "require_references": True
    }
)

Memory Persistence

from reactlm import RedisMemory

# Initialize memory system
memory = RedisMemory(
    host="localhost",
    port=6379,
    prefix="myapp:",
    ttl=3600  # 1 hour cache
)

# Use with agent
agent = Agent(
    llm=OpenAILLM(),
    config=AgentConfig(),
    memory=memory
)

📝 License

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