What makes Apptainer different from Docker in this skill?

Unlike Docker, Apptainer (formerly Singularity) is designed for HPC environments where users do not have root access. It uses single-file SIF images and maps the container user to the host user for better security on shared clusters.

Where are the container images stored?

By default, this skill saves pulled and built images as SIF files in the ~/.local/share/apptainer/ directory, keeping your workspace organized.

Can I build containers from scratch using this skill?

Yes, you can use the build action to create SIF images from standard Apptainer definition (.def) files, allowing for fully customized environment creation.

Can I pull images directly from Docker Hub?

Yes, you can pull images from Docker Hub, NVIDIA NGC, and other OCI-compliant registries by using the docker:// prefix in the image path.

How does the skill handle GPU acceleration?

The skill includes auto-detection logic that identifies your hardware. If an NVIDIA GPU is found, it automatically adds the --nv flag; for AMD, it adds --rocm, ensuring your container has access to the necessary drivers.

Apptainer HPC Container Manager

Name: Apptainer HPC Container Manager
Author: atrawog

byatrawog

Deployment & DevOps

Manages Apptainer (Singularity) containers for high-performance computing workloads with automatic GPU passthrough and SIF image support.

About

The Apptainer skill provides a comprehensive interface for managing Singularity-compatible containers specifically designed for High-Performance Computing (HPC) environments. It enables users to build, pull, and run SIF (Singularity Image Format) files without requiring root privileges, making it ideal for researchers and AI/ML engineers working on shared clusters. The skill automates complex tasks like GPU detection for NVIDIA and AMD hardware, definition file building, and cache management, ensuring that portable, reproducible environments are easily accessible within the Claude Code development workflow.

Key Features

Automatic GPU detection and passthrough for NVIDIA (CUDA) and AMD (ROCM) workloads
Rootless container execution compatible with strict HPC security requirements
Comprehensive cache management tools to list and clean Apptainer storage
Image pulling from multiple registries including Docker Hub, NVIDIA NGC, and Sylabs Cloud
SIF image management including building from definition files and metadata inspection
0 GitHub stars

Use Cases

Developing and testing AI/ML models in containers destined for HPC cluster environments
Creating portable, single-file reproducible research environments for data science
Running GPU-accelerated legacy or specialized software without root access on Linux systems

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add atrawog/overthink-plugins apptainer

For use in Claude.ai and ChatGPT

Download Skill

GitHub

About

Key Features

Automatic GPU detection and passthrough for NVIDIA (CUDA) and AMD (ROCM) workloads
Rootless container execution compatible with strict HPC security requirements
Comprehensive cache management tools to list and clean Apptainer storage
Image pulling from multiple registries including Docker Hub, NVIDIA NGC, and Sylabs Cloud
SIF image management including building from definition files and metadata inspection
0 GitHub stars

Use Cases

Developing and testing AI/ML models in containers destined for HPC cluster environments
Creating portable, single-file reproducible research environments for data science
Running GPU-accelerated legacy or specialized software without root access on Linux systems