AntonOsika/gpt-engineer
★ 55,024 stars
Covered in week 2026-W30 · the 2026-07 report
CLI platform to experiment with codegen. Precursor to: https://lovable.dev
What gpt-engineer is
By default, gpt-engineer expects text input via a prompt file. It can also accept image inputs for vision-capable models. This can be useful for adding UX or architecture diagrams as additional context for GPT Engineer. You can do this by specifying an image directory with the —-imagedirectory flag and setting a vision-capable model in the second CLI argument.
From the repository README · read it on GitHub →
Transcript
Writing boilerplate means copying the same imports, error handlers, and folder structures for every new project. You spend hours on syntax instead of actual logic. gpt-engineer turns plain English into working codebases. Describe what you want, it generates files, dependencies, and implementation. You ship features faster. Over 55,000 stars on GitHub show developers are tired of repetitive setup. Would you replace your boilerplate workflow with this?