flask_rag_admin

RAG-Powered Chat App with Admin Panel

Project Description

This is a full-stack chat application that mimics how an AI agent uses internal knowledge for context. It features:

• A React-based chat interface
• A Flask backend with a Retrieval-Augmented Generation (RAG) system
• A knowledge base built from structured Markdown (.md) files
• An integrated admin panel (Flask-Admin, Flask-RBAC) for user and knowledge management

Features

• Context-Aware Responses: RAG pipeline retrieves relevant information from Markdown files and uses it as context for LLM answers.
• Simple RAG System: Lightweight, keyword-based search over Markdown files (no vector DB required).
• Full-Stack Architecture: Flask + SQLAlchemy backend, React + Bootstrap frontend.
• Admin Panel: Manage users, roles, and trigger manual re-indexing of the RAG knowledge base.
• Authentication & Authorization: Flask-RBAC ensures only authenticated admins access privileged features.
• Folder-Based Knowledge: RAG indexes content from organized folders (resources/, docs/, etc.).
• Feedback System: Users can rate and comment on AI responses for quality improvement.
• User Settings: Each user can customize preferences (theme, notifications, etc.).
• Session Management: Secure user sessions and password reset functionality.

Technologies Used

Backend:

• Python 3.x
• Flask
• Flask-Admin
• Flask-RBAC
• SQLAlchemy
• Flask-SQLAlchemy
• Flask-Migrate
• SQLite (for the database)
• Ollama (for the LLM)

Frontend:

• React.js
• Bootstrap

Dev Tools:

• concurrently (to run backend and frontend together)

Other:

• Markdown for knowledge base files

Prerequisites

Before you begin, ensure you have:

• Python 3.x
• Node.js and npm
• Ollama running with an accessible LLM (e.g., llama3)

Backend Setup

cd backend
python -m venv venv
source venv/bin/activate  # On Windows, use venv\Scripts\activate
pip install -r requirements.txt

Frontend Setup

cd ../frontend
npm install
npm install concurrently --save-dev

Running Backend & Frontend Together

You can run both backend and frontend in development mode using concurrently:

1. Add the following script to your frontend/package.json:

   "scripts": {
     "dev": "concurrently \"cd ../backend && source venv/bin/activate && flask run\" \"npm start\""
   }

   (On Windows, replace source venv/bin/activate with venv\Scripts\activate.)

2. From the frontend folder, run:

   npm run dev

This will start both the Flask backend and React frontend together for development.

Usage

Access the chat interface:

• Navigate to http://localhost:3000 and log in.

Access the admin panel:

• Log in with an admin account and go to http://localhost:5000/admin
• Trigger RAG re-indexing after adding/modifying Markdown files
• Manage user accounts and roles

Roadmap

• Implement Semantic Search: Upgrade RAG pipeline to use semantic search with a vector database
• Expand Codebase Awareness: Index and understand code files (.py, .js, etc.)
• Add Built-in Commands: Integrate custom commands (e.g., @ mentions) into the chat interface
• Real-time Updates: Add live update feature using WebSockets

License

MIT