CodeGraph AI

An Autonomous AI DevOps Agent that visualizes, analyzes, and refactors legacy codebases.

CodeGraph is not just a chat wrapper. It is a RAG-powered Architect that:

1. Scans local Python codebases to build a Dependency Graph (AST Parsing).
2. Visualizes architecture using React Flow.
3. Refactors code autonomously with a Human-in-the-Loop approval pipeline.

🚀 Key Features

1. 🔍 AST-Based Context Awareness

Unlike standard chatbots that guess file relationships, CodeGraph uses Python's ast module to statically analyze imports and build a NetworkX directed graph. The AI understands exactly which files depend on main.py before suggesting changes.

2. 🛡️ Self-Healing AI Backend

The backend implements a Fault-Tolerant Model Router. It autonomously detects available Google Gemini models (Pro, Flash, Experimental) and dynamically switches to the highest-performance model available, ensuring 99.9% uptime even during API version deprecations.

3. ⚡ Autonomous Refactoring Pipeline

• Multi-File Editing: The agent can split large classes into separate files (schemas.py, utils.py) in a single pass.
• Safety First: Implements an automatic Rollback System. Every file edit creates a timestamped .bak backup before writing to disk.

🛠️ Tech Stack

• Frontend: React, React Flow, Glassmorphism UI (CSS Variables).
• Backend: FastAPI, Python NetworkX, Google Gemini 2.5 Flash.
• Analysis: Abstract Syntax Trees (AST) for static analysis.

📦 How to Run

1. Clone the repo

   git clone [https://github.com/Dhy4n-117/CodeGraph.git](https://github.com/Dhy4n-117/CodeGraph.git)
   cd CodeGraph

2. Backend Setup:

   cd backend
   pip install -r requirements.txt
   # Create a .env file with your GEMINI_API_KEY
   python main.py

3. Frontend Setup:

   cd frontend
   npm install
   npm run dev

4. Usage: * Open localhost:5173.

   • Enter the absolute path of any local Python project.
   • Click "Analyze Map".