There is no repo to connect
Publishing ·
Git-connected hosting assumes you have a repository, a framework, and someone who can wire a build. The people who produce the files we serve have an export. They do not have a deploy pipeline, and they should not need one.
The file is the input
Publidock takes a zip, a folder, or a single HTML file. Nested folders, spaces in names, an entry point that is not index.html — all handled. Relative paths stay intact. There is no framework detection because there is no framework.
Preview the exact bytes at a private address. Nothing is shared until you choose a rule. Re-publish to the same URL when the numbers change; roll back a bad version in one click.
When you do have a repo
A POST to /api/v1/sites/{slug}/versions is the same publish. Scoped tokens, a CI job, or an MCP client can ship the artifact without opening a browser. That is the secondary path. The primary path is the file that is already done.
How publishing works, including the messy-input case.