pubanon

Pubanon publishes selected files from a private Git repository to GitHub under a separate name and email. Each release creates one publication commit without copying your private commit history. Choose the files, review the prepared release, and confirm publication.

Pubanon checks the selected content for configured sensitive terms. It does not rewrite secrets out of files. You control what you publish; the content matching limits explain what the checks can detect.

Installation

You need Python 3.13 or newer, Git 2.41.0 or newer, and the GitHub CLI (gh). Pubanon targets Linux, macOS, and native Windows. See platform requirements for details.

From a checkout of this repository, install the command with uv:

uv tool install .

After updating the checkout, update the installed command:

uv tool install --reinstall .

uv is needed for these installation commands. If you use make, make install and make reinstall run the same commands.

Your first release

Choose your publication identity

pubanon setup

Setup asks for the name and email to use in published Git history and the terms you want to keep private. It verifies your GitHub CLI account and shows your answers before saving. Choose the name and email you want readers to see: existing publication commits keep that identity. Use pubanon setup --edit to change saved answers or pubanon doctor to check readiness.

Select your project and files

Enter a private source repository with no Git remote:

cd /path/to/private-project
pubanon init

Follow the prompts to choose files, a GitHub destination, and repository settings. Choose private visibility for this walkthrough so you can check the result on GitHub before making it public. Review your choices and save. Pubanon stores its configuration outside the source repository.

Prepare and review

Commit the changes you want to publish in your source repository, then run:

pubanon repo create
pubanon prepare
pubanon inspect

prepare saves the selected committed files as a local snapshot. Uncommitted changes are left out. If the selected pyproject.toml declares a package version, Pubanon uses it for the release tag; otherwise the snapshot is untagged. See version tags to choose an explicit version.

inspect shows the release details and a directory of read-only files to review. Builds or tests that write files need a separate writable copy. Changes in either copy do not alter the snapshot. To correct it, edit and commit in the source, then run pubanon discard and pubanon prepare again.

Publish and make it public

When you are satisfied, run:

pubanon publish
pubanon repo reveal

publish asks for confirmation before sending the saved snapshot to your private destination. repo reveal checks the published history, settings, and any configured CI requirements before asking to make the repository public. It also provides links to Actions logs and artifacts for you to review; Pubanon does not scan their contents.

For the next release, commit your changes and repeat prepare, inspect, and publish. For a versioned Python package, update and commit its version first. The publication guide also explains the release VERSION shortcut.

Still correcting the initial release privately? Commit your fixes, then use pubanon prepare --replace all, pubanon inspect, and pubanon publish to replace the entire published history while keeping the same version. To correct only the latest publication and keep earlier history, use pubanon prepare --replace. See correcting published history for the available scopes.

After replacements, pubanon clean previews and removes superseded local snapshots with one confirmation. It keeps current published snapshots and prepared work. See cleanup for other cleanup tasks.

Where to go next

Use pubanon status to see local state or pubanon status --remote to compare it with GitHub. If a push outcome is unknown, follow the reported recovery command: pubanon publish --resolve.

Development

Provide the tools in [dependency-groups].dev through your environment or package manager, then run:

make check

make format applies the repository formatters. Tests run offline with temporary Git repositories and a fake GitHub CLI.