mustafac0des/Multi-Document-RAG-Agent

👥 Role-Based Access Control RAG Agent

A powerful and secure AI chatbot that allows employees to query internal company documents with strict role-based access control.

✨ Features

• —🔐 Role-Based Access Control (RBAC) - Users only see information they are authorized to access based on their simulated role
• —📚 Multi-Document Ingestion - Processes Markdown files and extracts valuable metadata for precise querying
• —🧠 Local LLM & Embeddings - Fully local generation and embeddings using open-source HuggingFace models for data privacy
• —🗃️ Vector Database - Utilizes ChromaDB for fast, efficient, and persistent document retrieval
• —💬 Gradio Interface - An easy-to-use, interactive web UI for chatting with the AI

🚀 Getting Started

Installation

1. 1.Clone the repository
2. 2.Install the required dependencies: pip install gradio transformers sentence-transformers langchain-core langchain-chroma pyyaml python-dotenv torch
3. 3.Create or configure your .env file in the root directory

Running the App

1. 1.Open your terminal in the project folder
2. 2.Run the application: python app.py
3. 3.Access the web interface via the local URL provided (usually http://127.0.0.1:7860)

🎯 Quick Tips

• —Use the role dropdown to test how different roles (e.g., HR, Executive, Engineer) affect the AI's responses
• —Ensure your Markdown documents in the company_documents folder have correct YAML frontmatter for roles
• —Ask specific, detailed questions for the best response from the LLM
• —If the AI says "I cannot help you with that!", try changing your simulated role to one with higher clearance

❓ Questions?

• —Can I add new documents? - Yes, place new Markdown files in the company_documents folder and restart the app
• —How do I change roles? - Select a different role from the dropdown menu in the Gradio UI
• —Is my data sent to the cloud? - No, the app uses local embeddings and a local LLM (google/flan-t5-base) to keep data secure
• —Why is the AI not answering my question? - You either lack the required role to view the source document, or the answer isn't in the knowledge base

🛠️ Tech Stack

• —UI Framework: Gradio
• —Vector Store: ChromaDB
• —Embeddings: SentenceTransformers (all-MiniLM-L6-v2)
• —LLM: HuggingFace Transformers (google/flan-t5-base)
• —Orchestration: Langchain Core