A plain git repository

A repository on Pando does not need projects. Any ordinary git repo works on its own.

cd ~/code/acme          # any existing git repo
pando init              # it becomes a Pando repo, ramos/acme

pando init pushes your default branch as main and changes nothing else in the checkout. It adds worktrees/ and .pando/local/ to the repo’s .git/info/exclude, so git ignores them.

On another machine:

pando clone ramos/acme

From then on, the repo looks like this on every machine:

acme/
  worktrees/
    main/          main, read-only, follows every merge
    fix-login/     a branch, one agent

What you get

Everything Pando gives code:

• agents working side by side, each on its own branch in its own worktree;
• review of a branch’s diff and commits, and Merge to land it;
• checks run by Buildkite or any CI, with Merge waiting for the ones you require (see Checks and CI);
• a main nobody can push to directly.

Plain git clone, git pull and git push against the repo’s address (https://<server>/<owner>/<name>.git) work too. git asks for a password: give it a credential from pando credential create.

Settings that a repository with projects keeps in its shared .pando/pando.toml live on this repo’s main branch instead, at .pando/pando.toml, so they change through a reviewed branch like any other file. Required checks go under [repos."."].

What you do not get

The shared files around the projects, and references. A plain repository is one git repo, with nothing around it.

How Pando tells the difference

pando init in a folder that is a git repo makes a plain repository. In a folder that is not, it makes a repository with projects: each git repo under projects/ becomes one of its projects.