QuantaLogic is a Python framework for building AI agents that can code, automate workflows, and engage in natural conversations, powered by various LLMs. It targets developers, business innovators, and AI enthusiasts seeking flexible and efficient AI solutions, offering modes for dynamic task-solving (ReAct), structured automation (Flow), and conversational interaction (Chat).

How It Works

QuantaLogic leverages LLMs through the LiteLLM library, supporting providers like OpenAI, Anthropic, and DeepSeek. Its core strength lies in its versatile modes: ReAct for iterative problem-solving using tools and code, Flow for building sequential or parallel workflows via a node-based system, and Chat for interactive conversations with integrated tool usage. A key innovation is CodeAct, an extension that prioritizes executable Python code as the agent's primary action language for complex, multi-step tasks.

Quick Start & Requirements

• Install: pip install quantalogic and optionally pip install quantalogic-codeact.
• Prerequisites: Python >= 3.10. Docker is recommended for sandboxed execution.
• Configuration: API keys for LLMs should be placed in a .env file.
• Docs: Full Docs, Quick Guide

Highlighted Details

• Supports three distinct modes: ReAct, Flow, and Chat.
• CodeAct extension enables agents to use executable Python code as their primary action language.
• Integrates with multiple LLM providers via LiteLLM, including local models via LM_STUDIO_API_BASE.
• Offers a CLI for direct interaction and a Python API for programmatic use.
• Includes built-in tools for coding, web search, and file management, with support for custom tools.

Maintenance & Community

• The project is actively maintained by Raphaël MANSUY.
• Contributions are encouraged via fork, branch, code, and PR. See CONTRIBUTING.md.

Licensing & Compatibility

• Licensed under Apache 2.0, permitting commercial use and modification.

Limitations & Caveats

• Docker is optional but recommended for secure execution, implying potential security considerations for direct execution without it.
• The README mentions "CodeAct vs ReAct: What's the Difference?" and links to "CodeAct Module" for detailed documentation, suggesting CodeAct might be a newer or more complex component.