A "ReAct" (Reason + Act) Agent built from scratch in Python without external LLM libraries.
This project demonstrates the core concept of Agentic AIโmoving beyond chatbots to systems that can autonomously execute tools.
Unlike a standard LLM which only outputs text, this agent:
- Reasons about a user request.
- Decides which tool to use (Calculator or Security Scanner).
- Executes the tool in a safe Python environment.
- Returns the actual result to the user.
- Mock LLM Brain: Simulates decision-making logic using Regex patterns.
- Tools:
calculate(): Safely evaluates mathematical expressions.check_ip_reputation(): Simulates a cybersecurity scan for malicious IPs.
- Safety: Implements input sanitization and error handling.
- Clone the repository:
git clone [https://github.com/krinathakkar646/Simple-Agentic-AI.git](https://github.com/krinathakkar646/Simple-Agentic-AI.git)
- Run the Agent
python agent.py