Not open yet — signups haven't started. Join the list for the day they do.
Home/Documentation/Connect a Git repository

Documentation

Connect a Git repository

Connect a repository once. After that, every push to the branch you choose republishes the site — no dashboard visit, and no build step running on our side.

Quick factsthe numbers
99¢a month
1 GBstorage
250 GBtraffic
99%uptime target

Connecting a Git repository links one repository to one site. From then on, a normal git push to the branch you picked is the entire deploy process — nothing else to run, click, or remember. This page covers connecting and publishing in plain terms; see the Git deploy page for the fuller walkthrough.

Connect once, publish on every push

Authorize access once, from your account, and pick the repository. That's the only step you'll repeat less than daily — from then on, pushing to your chosen branch is the deploy. There's no separate "deploy" button to click afterward; the push does it.

Each push joins a queue and publishes in order, so several commits landing close together don't publish out of sequence, and a wave of pushes across many sites never makes one site's deploy wait behind an unrelated one.

Branch and output folder

Two settings control what actually gets published:

SettingWhat it controlsTypical value
BranchWhich branch triggers a deploy when you push to itmain
Output folderWhich folder in the repo gets published — the whole repo, or a subfolderrepo root, or /public, /dist

If your project builds into a subfolder — a static site generator that outputs to /public, a bundler that outputs to /dist — point the output folder there rather than at the repo root. Only the files inside that folder get published; everything else in the repository, including source files, configuration, and a README, stays out of the live site.

No build step runs on our side

We don't run a build on our side. Whatever's sitting in the folder you pointed at is exactly what gets served — no build command to configure, no dependency that fails to install on our end, no framework version drifting out of sync with your local machine.

In practice, that means:

  • Run your build wherever you already do — locally, or in your own CI — and push the built output
  • If your generator or bundler needs a build command, run it before you push, not after
  • Nothing here fails because a build broke, because no build runs here

Two things that trip people up

index.html at the wrong level. If your homepage loads blank or returns a 404, check that index.html sits directly inside the folder you pointed at — not nested one level down inside a subfolder that folder happens to contain.

Case-sensitive paths. A link to /About.html won't find about.html. It can work on a local machine and break here, because the underlying storage treats the two as different files. Match case exactly, in every link and file name.

Keep reading

Other ways to get files onto a live URL

Get on the list

Join the list and we'll email you just once — that's the only thing we use your address for.

Join the list