README.md

metadata

title: Code Review Agent Environment
sdk: docker
app_port: 7860

Code Review Agent Environment

Submission for the Meta x Hugging Face OpenEnv Hackathon.

Authors / Members:

• Ashish
• Shardul
• Harshit

Overview

This project is a real-world code review environment for AI agents. The agent receives a code diff plus surrounding context and must identify issues, suggest fixes, and decide whether changes should be approved or rejected.

Live Section

Live Space: huggingface.co/spaces/Spirit-26/code-review-environment

Core Features

• OpenEnv-style reset(), step(), and state() API
• Pydantic-based action, observation, and state models
• 6 review tasks across easy, medium, and hard difficulty
• Deterministic graders with scores from 0.0 to 1.0
• Reward shaping for partial progress and invalid actions
• Docker support and Hugging Face Space deployment
• Baseline inference script at the repository root: inference.py

Required Environment Variables

Use an OpenAI-compatible endpoint.

• API_BASE_URL
• MODEL_NAME
• API_KEY

Backward-compatible aliases are also supported:

• HF_TOKEN
• OPENAI_API_KEY

Local Run

python -m venv .venv
.\.venv\Scripts\Activate.ps1
pip install -r requirements.txt
python inference.py --task-id bug_detection_easy_1 --max-steps 50 --output baseline_results.json

Validation

python submit.py

Docker

docker build -t code-review-openenv .
docker run --rm -p 7860:7860 \
  -e API_BASE_URL=https://api.openai.com/v1 \
  -e MODEL_NAME=gpt-4o-mini \
  -e API_KEY=<your_key> \
  code-review-openenv