podman-minimal

Clean environments. One command.

Launch a disposable Podman shell from the folder you are already in. Current directory is mounted by default, your UID/GID is preserved, and Dockerfile/devcontainer layouts are auto-discovered.

custom CUDA imageGPU visible inside the container

$ podman-minimal --image docker.io/nvidia/cuda:12.4.1-base-ubuntu22.04 nvidia-smi -L

GPU 0: NVIDIA GeForce RTX 5080

Same command shape works for CPU images, local Dockerfiles, one-off jobs, and long-running services.

Install

python3 -c "import urllib.request as u; exec(u.urlopen('https://raw.githubusercontent.com/vincenzoml/podman-minimal/refs/heads/main/podman-minimal.py').read().decode())" --install

Auto mode detects your OS. Docs: README.

Interactive workspace

Drop into a clean shell with your current project mounted, without installing toolchains on the host.

Batch commands

Run tests, scripts, GPU probes, or one-shot jobs in the same reproducible container shape.

Persistent work

Use --nohup for terminal-proof jobs or Linux-only user services for daemons.

Why not plain `podman run`?

No-hassle bootstrap: one command installs the launcher and prepares Podman across Linux, macOS, and Windows.
No-hassle identity and workspace defaults: runs as your same UID/GID with the current working directory already mounted.
Automatic image workflow: detects Dockerfiles, builds/reuses tagged images, or pulls when needed.
VS Code-friendly defaults: handles `.devcontainer` / `.devcontainers` Dockerfile locations and correct build context automatically.
Built-in quality-of-life modes: quiet by default, optional verbose diagnostics, detached `--nohup`, and Linux daemon helpers.

podman-minimal is free software. If it saves you time, please star the repository: it is the easiest way to say thanks and helps other developers find it.