README.md

metadata

title: OpenEnv Python Debugger
emoji: 🐍
colorFrom: blue
colorTo: yellow
sdk: docker
app_port: 7860
tags:
  - openenv
  - python
  - debugging

OpenEnv Python Debugger

Status: Deployment triggered.

A real-world OpenEnv environment where an AI agent learns to write, debug, and optimize Python code by working with a live Python interpreter.

The Environment

This environment simulates a core task software engineers do daily: taking broken code and making it pass tests.

Action Space: The agent outputs raw Python code:

{
  "code": "def my_func():\\n    return True"
}

Observation Space: The environment responds with test execution details:

{
  "task_id": "task_easy",
  "current_code": "def my_func():\\n    return False",
  "test_results": {
    "success": false,
    "tests_passed": 0,
    "total_tests": 1,
    "error": "AssertionError at test_my_func",
    "stdout": ""
  },
  "feedback": "Tests failed:\\nAssertionError at test_my_func",
  "step": 1,
  "max_steps": 10,
  "score": 0.0
}

Tasks

1. Python Syntax Fix (Easy): Fix straightforward syntax and semantic errors.
2. Python Logic Repair (Medium): Fix algorithmic bugs in correctly formatted code.
3. Algorithm Optimization (Hard): Optimize a slow O(N^2) algorithm into an O(N) one to bypass timeout tests.

Local Setup

1. Install dependencies:

pip install -r requirements.txt

2. Run the environment:

uvicorn app:app --port 7860 --host 0.0.0.0

3. Run the baseline:

export API_BASE_URL="your-llm-base-url"
export API_KEY="your-api-key"
export MODEL_NAME="your-model-name"
python inference.py

Validation

This environment passes openenv validate out of the box.