usmo 0.1.5

Bootstrap machines and run cached utility scripts from one CLI.

usm

usm is a lightweight CLI for machine bootstrap tasks and day-to-day utility scripts. It gives you one command for common setup jobs, Azure/blobfuse workflows, and a few personal productivity helpers, while keeping the underlying scripts easy to iterate on.

The package installs as usmo, but the executable command is usm.

Highlights

• One entrypoint for machine setup, storage helpers, and quick admin tasks.
• On-demand script download and caching under ~/.cache/usm/scripts.
• Python subcommands run with the package interpreter, which keeps pipx installs isolated and reliable.
• Local --debug mode for iterating on scripts in this repository without downloading from GitHub.
• Simple release flow driven by Git tags.

Installation

Install from PyPI with pipx:

pipx install usmo

PyPI package page: https://pypi.org/project/usmo/

Or install from a local checkout while developing:

uv sync
pipx install --force .

After installation, the CLI is available as:

usm <command> [args...]

Commands

Command	Description
usm init	Bootstrap a fresh Ubuntu machine with common packages, shell aliases, pipx tools, tmux plugins, and Neovim config.
usm blobmount <mount_dir> <account> <container>	Install blobfuse2 if needed, generate a SAS token from your Azure CLI login, and mount a blob container locally.
usm cu122	Install NVIDIA driver 535, CUDA 12.2, and Vulkan-related packages on Ubuntu.
usm cp [--use-sas-token] <source>... <destination>	Copy between local paths and blobfuse2 mountpoints, delegating to azcopy when Azure storage is involved.
usm check_py	Print the active Python and pip locations and versions.
`usm inject-alias [--shell bash	zsh

Examples

Initialize a new machine:

usm init

Mount a blob container:

usm blobmount /mnt/data myaccount mycontainer

Copy from a blobfuse mount to a local directory:

usm cp /mnt/data/project ./project-backup

Refresh the cached script before running it:

usm --upgrade check_py

Run against the local scripts/ directory instead of downloading from GitHub:

usm --debug check_py

Inject aliases into your zsh profile:

usm inject-alias --shell zsh

Inject aliases into your PowerShell profile:

usm inject-alias --shell powershell

Write the managed alias block into a specific file:

usm inject-alias --file ~/.config/usm/test-shell.rc

Write PowerShell-flavored aliases into a custom profile file:

usm inject-alias --shell powershell --file ~/Documents/PowerShell/Microsoft.PowerShell_profile.ps1

How it Works

The CLI keeps a small manifest of commands in src/usmo/cli.py.

• Shell scripts are executed with bash.
• Python scripts are executed with the current interpreter via sys.executable.
• Remote scripts are downloaded from this repository and cached locally.
• --upgrade forces a fresh download of the selected script.
• --debug bypasses the cache and runs the local file under scripts/.
• Managed alias insertion uses start/end markers so rerunning the command updates the block instead of duplicating it.
• inject-alias is implemented with click and supports bash, zsh, and PowerShell profile targets.

Development

Install the project locally:

uv sync

Build distributable artifacts:

uv build

Smoke-test the installed command:

uv run usm check_py

Release Flow

This repository includes a GitHub Actions workflow at .github/workflows/release.yml.

When you push a tag like v0.1.4, GitHub Actions will:

1. Build the source distribution and wheel with uv build.
2. Create or update a GitHub Release for that tag.
3. Upload the built artifacts to the release page.
4. Publish to PyPI if the repository secret PYPI_API_TOKEN is configured.

Create a new release with:

git tag -a v0.1.4 -m "v0.1.4"
git push origin v0.1.4

Notes

• Some scripts are tailored for Ubuntu-based environments.
• blobmount and cp expect Azure CLI / azcopy / blobfuse2 style workflows.
• Cached scripts live in ~/.cache/usm/scripts.