1. Fill in a form
No git, no terminal, no code. Answer a structured web form and a maintainer turns it into a pull request for you. This is the recommended path for most people.
Contribute
Contribution is membership
There is no signup form. The moment your first pull request merges, you're a member — your name appears on the contributors page and you get a vote in governance. Three ways in, and only one of them needs git.
2. Edit in the browser
Every page has a "Suggest an edit" link at the bottom. It opens GitHub's web editor, forks the repo for you, and opens a pull request when you save. Fixing a typo takes 30 seconds.
3. Clone and PR
The developer path. Fork, branch, add a card or course under src/content/, run npm run build, open a PR. CI validates your schema, licenses, and links automatically.
What happens to your submission
- Schema validation — required fields, valid dates, well-formed structure.
- License check — the SPDX allowlist rejects restrictive "open-washing" licenses mechanically.
- Link & checksum check — every artifact URL must resolve; checksums must be well-formed.
- Build & preview — the site builds with your change so reviewers see it rendered.
- Human review — a maintainer responds. Our target is under 48 hours; we publish the actual number.
- Merge → live — production deploys automatically, usually within two minutes.
What we're looking for right now
Genuinely open models
Open weights and documented training data and a permissive license. If any axis is closed, we still list it — clearly graded, so nobody is misled.
Ethically sourced datasets
You'll need to state collection method, consent basis, PII review status, and known bias. These are required fields — the schema refuses cards without them.
Courses in any language
Especially non-English. AI education that only exists in English is not equitable AI education.
Translations and accessibility fixes
Unglamorous, disproportionately valuable, and always merged fast.