01 · section
Authenticate first
Publishing requires a GitHub-backed login flow because namespace ownership is tied to your Registry identity.
Login
$ axint loginGetting started
Publish to the Axint Registry.
Publishing today is JSON-first: the CLI validates your package, checks namespace ownership, stores metadata in D1, stores the large payload in R2, and returns the live package URL.
02 · section
Dry run, then publish
Treat dry-run as part of the normal publish loop. It is the best way to catch missing metadata before a bad version lands publicly.
- The namespace must match your GitHub-linked registry user, unless you are publishing to an org-controlled namespace like @axint.
- The current production API accepts JSON payloads, not tarball uploads.
- The publish response returns the live package URL.
Publish flow
$ axint publish --dry-run
$ axint publish03 · section
What gets stored
D1 stores searchable package and version metadata. R2 stores the large payload for source, Swift output, plist fragments, and IR.
- Package metadata: name, namespace, slug, description, latest version, owner
- Version metadata: compiler version, language, tags, surface areas, README
- Large payload: source, Swift output, plist fragment, IR