A RAG-based Code Debugger built using Streamlit, LangChain, ChromaDB, and Ollama (Phi-3). Upload Python code or PDFs and ask questions to identify bugs, understand logic, and get fix suggestions. Fully local, no API keys required, and beginner-friendly. This project runs completely locally using Ollama (Phi-3), so no API keys are required. It combines document retrieval using ChromaDB with intelligent responses generated by an LLM. The interface is built using Streamlit, making it simple and interactive.
Key Features
Upload .py, .txt, or .pdf files
Ask natural language questions about the code
Automatic bug detection and explanations
Uses local LLM (Phi-3) via Ollama
Clean and simple Streamlit UI
Download generated answers Follow these steps to run the RAG Code Debugger on your local machine:
Open the project folder in VS Code
Activate the virtual environment
source venv/bin/activate
Install required dependencies
pip install -r requirements.txt
Ensure Ollama is running and the Phi-3 model is available
ollama list
Start the Streamlit app
streamlit run main.py
A browser window will open automatically.
Upload a Python (.py), text (.txt), or PDF (.pdf) file
Enter a question about the code
View the AI-generated explanation or fix
The application runs fully offline using Ollama, and no API keys are required.