The docker system prune -af --volumes command removes all Docker build cache, which may slow down subsequent Docker builds by preventing layer reuse. This action runs before Docker build steps in both workflows, so cached layers from previous builds that could speed up the current build will be deleted.

Consider either:

1. Removing the --volumes flag to preserve named volumes
2. Using a more selective prune command like docker builder prune -af --filter "until=24h" to keep recent cache
3. Moving this step to run after Docker builds complete, or making it conditional
4. Adding error handling with || true to make this step best-effort rather than failing the job

The cleanup steps in this action have no error handling. If any command fails (e.g., a directory doesn't exist or Docker is not available), the entire job will fail. Since freeing disk space is typically a best-effort operation to prevent failures rather than being critical itself, consider adding || true to each command or using set +e to make failures non-fatal.

For example:

sudo rm -rf /usr/share/dotnet || true
sudo rm -rf /usr/local/lib/android || true

This ensures that if one cleanup operation fails, the others can still proceed and the main job continues.