AutoCodeAI

Project Overview

This system demonstrates an autonomous AI development workflow that transforms natural language prompts into fully functional Python applications. The toolkit features:

• Single-process version (singleprocess.py) for straightforward execution
• Multi-process version (multiprocess.py) with parallel processing capabilities
• Full automation of coding, testing, debugging, and documentation

How It Works

Core Workflow

graph LR
    A[User Prompt] --> B(Prompt Optimization)
    B --> C(Code Generation)
    C --> D[Environment Setup]
    D --> E{Execution}
    E -->|Success| F[Documentation]
    E -->|Failure| G[AI Debugging]
    G --> C

Loading

Key Components

1. Prompt Refining: Enhances user input for better AI comprehension
2. Code Generation: Creates Python code + dependency installation commands
3. Isolated Execution: Runs code in dedicated virtual environments
4. AI Debugging: Automatically fixes errors through iterative improvements
5. Documentation: Generates GitHub-ready README files

Setup Instructions

Prerequisites

# Install required packages
pip install openai

Configuration

1. Replace "API-KEY" in both scripts with your DeepSeek API key
2. Ensure Zsh is installed (sudo apt install zsh for Linux)

Execution

Single-process version

python singleprocess.py

Multi-process version (3 parallel attempts)

python multiprocess.py

Output Structure

Successful runs create:

random_folder/
├── venv/                 # Virtual environment
├── generated_code.py     # Functional Python code
├── README.md             # Project documentation
└── log.txt               # Execution logs with error codes

Important Notes

1. System Compatibility: Designed for Unix-like systems (macOS/Linux)
2. Error Codes:
   • 200: Successful execution
   • 400: Runtime error (triggers debugging)
   • 5000: Critical failure
3. Safety Features:
   • Limited to 3 debugging iterations
   • Isolated virtual environments
   • Complete output logging

Example Output

# Sample generated code (simplified)
import numpy as np

def calculate_stats(data):
    return {
        "mean": np.mean(data),
        "median": np.median(data),
        "std_dev": np.std(data)
    }

# Sample dependency installation
pip install numpy