🔧 CodeFixer

An OpenEnv-based Reinforcement Learning environment designed to train AI agents to debug Python code automatically.

This project focuses on creating a structured environment where models learn to fix buggy code using feedback-driven rewards.

---

🚀 Overview

CodeFixer provides an interactive environment where an AI agent:

• Receives buggy Python functions
• Understands the intended functionality
• Submits corrected code
• Gets evaluated through automated test cases
• Learns using a reward-based feedback system

This setup mimics real-world debugging scenarios and is ideal for training RL-based code assistants.

---

🧠 Key Features

• ✅ Reinforcement Learning environment (OpenEnv compatible)
• ✅ Automated grading using deterministic test cases
• ✅ Dense reward signals for better learning
• ✅ Modular architecture for adding new tasks
• ✅ Docker support for easy deployment

---

🛠️ Tech Stack

• Python
• OpenEnv
• FastAPI
• Pytest
• Docker

---

🏗️ Project Structure

CodeFixer/
├── env/
│   ├── environment.py     # Core RL environment logic
│   ├── models.py          # Data models (Observation, Action, Reward)
│
├── tasks/
│   ├── tasks.py           # Buggy coding tasks & test cases
│
├── tests/
│   ├── test_env.py        # Unit tests for environment
│
├── inference.py           # Inference / evaluation script
├── server.py              # FastAPI server
├── requirements.txt       # Python dependencies
├── Dockerfile             # Container setup
├── openenv.yaml           # OpenEnv configuration

---

⚙️ Installation

1. Clone the repository

git clone https://github.com/Swaroop-Haridas/CodeFixer.git
cd CodeFixer

2. Create virtual environment

python -m venv venv
source venv/bin/activate   # Linux/Mac
venv\\Scripts\\activate      # Windows

3. Install dependencies

pip install -r requirements.txt

---

🐳 Docker Setup (Recommended)

docker build -t codefixer .
docker run -p 8000:8000 codefixer

---

▶️ Usage

Run Environment Locally

python inference.py

Run Server

python server.py

---

🔄 How It Works

1. Environment provides:

   • Buggy code
   • Problem description

2. Agent submits:

   • Fixed Python function

3. Environment evaluates:

   • Correctness (test cases)
   • Efficiency
   • Code validity

4. Reward is computed and returned

---

🧪 Testing

Run unit tests using:

pytest

---

📌 Use Cases

• Training RL agents for code debugging
• Building AI coding assistants
• Research in program synthesis & repair
• Educational tools for learning debugging

---

"Teaching machines to debug code is one step closer to autonomous programming." 🚀