gpt-engineer
A coding assistant CLI that writes and runs code from plain-language requests. Runs locally or in Docker, with local models or OpenAI and Anthropic APIs.
gpt-engineer is a locally run coding assistant for developers who want to experiment with AI code generation and build their own agents. It can create software from plain-language descriptions or make requested changes to an existing codebase. The project is archived and no longer maintained. Its Python code is open source under the MIT license.
The assistant writes and executes code in a project folder on your computer. It supports Windows and can also run in Docker. Model choice determines where inference happens: it supports local models such as WizardCoder, alongside OpenAI, Azure OpenAI and Anthropic APIs. Using those APIs sends model requests to the provider, so running the CLI locally doesn't mean the AI processing stays on your hardware.
For experiments beyond text prompts, vision-capable models can accept images, including interface sketches or architecture diagrams, as context for code generation. Custom agent prompts let developers change the assistant's behavior and carry shared instructions between projects. These controls make it relevant to people testing how coding agents work, alongside those generating an initial codebase.
It also includes benchmarking for custom agents against the APPS and MBPP datasets.
