Configure GHCR for HA add-on image publishing

[cayossarian]

Context

The HA add-on build pipeline (.github/workflows/build-addon.yaml) builds multi-arch Docker images and pushes them to GHCR at ghcr.io/spanpanel/simulator/{arch}:{version}. The HA Supervisor pulls these images when a user installs the add-on.

The workflow triggers on pushes to main but will fail until GHCR permissions are configured.

Required Setup

1. Enable GitHub Packages for the organization

In the SpanPanel org settings:

• Settings > Packages > Package creation — ensure members can publish packages

2. Grant the repository write access to packages

The workflow uses GITHUB_TOKEN with packages: write permission. This requires:

• Repo Settings > Actions > General > Workflow permissions — set to "Read and write permissions"

3. Verify the first image push

After the above is configured:

• Manually trigger the workflow via Actions > Build HA Add-on Image > Run workflow (or push any change to main that touches a path in the trigger list)
• Verify images appear at https://github.com/orgs/SpanPanel/packages
• Two packages should be created:
  • simulator/aarch64 (ARM64)
  • simulator/amd64 (x86-64)

4. Make packages public

Since the repo is public, GHCR packages should be public so HA Supervisor can pull without authentication:

• Go to the package settings page for each package
• Verify visibility is Public

Verification

Once images are published, test the full add-on install flow:

1. In HA, go to Settings > Add-ons > Add-on Store > Repositories
2. Add https://github.com/SpanPanel/simulator
3. SPAN Panel Simulator should appear in the store
4. Install it — the Supervisor should pull the image for the correct architecture
5. Configure options and start — verify the simulator runs and the span-panel integration discovers it