Cheralia/Ethiopia-Commercial-Code-Assistant

# ⚖️ Ethiopia Commercial Code AI Agent **A Self-Reflective RAG Assistant built with LangGraph, Streamlit, and Ollama.** Welcome to the **Ethiopia Commercial Code Assistant**! This isn't just a simple chatbot; it's an intelligent **agent** that "thinks" before it speaks. By using **Self-Reflection**, it critiques its own answers, checks for hallucinations, and ensures every response is grounded in the official Commercial Code of Ethiopia Proclamation. --- ## 🚀 Key Features * **🧠 Self-Correction**: If the agent hallucinates or gives a vague answer, it catches itself and retries. * **🔍 Smart Retrieval**: Uses Vector Search (FAISS) to find exact legal articles. * **⚡ Hybrid Intelligence**: Uses a powerful cloud model (GPT-4o/Groq) for drafting answers and a fast local model (**Qwen 2.5**) for critiquing them (saving you money!). * **🔄 Query Transformation**: If your question is unclear, the agent rewrites it to find better legal matches. * **📊 Interactive UI**: A clean, professional interface built with Streamlit. --- ## 📂 Project Structure ```text ethiopia-commercial-code-agent/ │ ├── assets/ │ └── architecture_final.jpeg │ └── graph.png │ ├── data/ │ └── commercial_code.pdf # 📜 Place your Proclamation PDF here │ ├── vectorstore/ # 🗄️ Vector Database (Auto-generated) │ └── faiss_index/ │ ├── src/ │ ├── __init__.py │ ├── config.py # ⚙️ Configuration & Model Selection │ ├── ingestion.py # 📥 PDF Processing & Embedding │ ├── graph.py # 🕸️ LangGraph Logic (The "Brain") │ ├── chains.py # 🔗 Prompts for Generation & Critique │ └── cache.py # 🚀 Caching Mechanism │ ├── app.py # 🖥️ Streamlit Frontend ├── requirements.txt # 📦 Python Dependencies ├── .gitignore ├── .env # 🔑 API Keys └── README.md ``` --- ## 🛠️ Installation & Setup Follow these steps to get the agent running on your machine. …