Readme-AI is an open-source command-line tool, built in Python, that scans a repository's file structure and source code, then writes a full README.md from the result. It can call OpenAI, Anthropic, Gemini, or a local Ollama model to draft the prose, or run in an offline mode that skips the API calls and produces boilerplate sections instead.
The output includes a table of contents, badges, installation instructions, usage examples, and a project overview, with options to customize the template, add a logo, and control emoji use. Developers install it with pip, run it through Docker, or clone the repository directly, and a .readmeaiignore file lets them exclude specific paths from the scan.
The project has around 2,300 GitHub stars and 220 forks, with a Streamlit-based web UI for people who would rather not use the CLI. It fits a narrow but genuinely useful niche: writing the README nobody gets around to writing properly.