# GPT Engineer — Generate Entire Codebases from Prompts

> CLI tool that generates entire codebases from natural language prompts. Iterative refinement with human feedback. Supports Python, JS, and more.

## Install

Save as a script file and run:

## Quick Use

```bash
pip install gpt-engineer

# Create a project folder with a "prompt" file describing what you want
mkdir my-project && cd my-project
echo "Create a snake game with Python and Pygame" > prompt

# Generate the code
gpte .
```

**Required**: `export OPENAI_API_KEY=your_key`

---

## Intro

GPT Engineer lets you describe software in plain English and watch AI write the entire codebase. Point it at a folder with a `prompt` file, and it generates a complete, runnable project. You can also use it to improve existing code by running `gpte . -i`. With 55,000+ GitHub stars, it's one of the original AI code generation experiments.

> **Note**: The team now focuses on [gptengineer.app](https://gptengineer.app) (managed service). For CLI AI coding, the authors recommend [Aider](https://github.com/Aider-AI/aider) as a well-maintained alternative.

**Works with**: Claude Code, GitHub Copilot, OpenAI Codex
---

## Usage

### Create new code
```bash
mkdir my-project && cd my-project
echo "Build a REST API with Flask that manages a todo list" > prompt
gpte .
```

### Improve existing code
```bash
cd existing-project
echo "Add input validation and error handling to all endpoints" > prompt
gpte . -i
```

### Custom model support
Supports OpenAI, Azure OpenAI, Anthropic models, and open-source alternatives like WizardCoder.

### Vision capability
```bash
gpte projects/my-app gpt-4-vision-preview \
  --prompt_file prompt/text \
  --image_directory prompt/images -i
```

### Pre-prompts
Customize AI behavior with `--use-custom-preprompts` for consistent coding patterns across projects.

## Installation

**Stable**: `pip install gpt-engineer`
**Development**: Clone repo → `poetry install` → `poetry shell`
**Docker**: Available for containerized usage
**Codespaces**: One-click browser-based setup

**Python**: 3.10 - 3.12 required

## Benchmarking

Built-in `bench` tool for testing custom agents against APPS and MBPP datasets:
```bash
bench my-agent --dataset apps
```

---



### FAQ

**Q: What is GPT Engineer?**
A: CLI tool that generates entire codebases from natural language prompts. Iterative refinement with human feedback. Supports Python, JS, and more.

**Q: How do I install GPT Engineer?**
A: Check the Quick Use section above for step-by-step installation instructions. Most assets can be set up in under 2 minutes.

## Source & Thanks

> Created by [Anton Osika](https://github.com/AntonOsika). Licensed under MIT.
> [gpt-engineer](https://github.com/AntonOsika/gpt-engineer) — ⭐ 55,200+

Thanks to Anton Osika and the community for pioneering prompt-to-codebase generation.

---
Source: https://tokrepo.com/en/workflows/84f451e1-0302-40e1-8dbd-f1942050a277
Author: Script Depot