OSAI Atlas
Sign in
Methodology

Where every number comes from.

Useful open-source AI, explained clearly. Here is how we present project information and maintenance signals.

Project directory

Open-source products organized by purpose, with direct links to their repositories, websites and documentation.

Directory

Stars, forks, licence, language, push date, releases

GitHub REST API, fetched per repository. If a request fails or the repo isn't on GitHub, the value is shown as unknown.

GitHub

Trending

Star gain measured between our own syncs. If there is not enough history, we show popular projects instead and label them as such.

Derived

Activity

A coarse label from the last push date. It's a signal, not a judgement of quality.

Derived

Guides

Drafted by a language model from the README and metadata, marked as such, and editable by a human. Edited guides are labelled.

AI + editor

What we won't do

  • Invent a metric. Missing data reads "unknown".
  • Accept payment for placement. Editor's picks are editorial.
  • Present AI-drafted prose as reviewed when it isn't.

Project ownership

Each product belongs to its respective maintainers. Follow the product’s own repository, license and documentation for authoritative information, installation instructions and support.