sue-slurm-web-design

What problem does it solve?

This Skill solves the challenge of monitoring Slurm HPC job status, compute-hour budgets, and storage usage across firewalled sandbox backends (NM5, LUMI, Snellius, Brev) where direct Slurm-web deployment is not feasible due to network restrictions, by providing a custom Vercel-hosted dashboard that reuses Slurm-web's visual design language.

Core Features & Use Cases

• End-to-end design-to-deploy workflow: Covers the full arc from dashboard design and data model definition to scaffolding the Next.js + Turso frontend, building the SSH collector, and deploying to Vercel with proper access controls.
• Multi-backend support: Works with standard Slurm backends (NM5, LUMI, Snellius) and non-Slurm Brev/DGX Cloud backends, with collector logic adapted per backend while reusing the shared frontend and database schema.
• Accurate metric tracking: Enforces strict unit conversion discipline for compute-hour budgets (including MN5's CPU-core-hour to GPU-hour conversion) and storage usage, with redaction of sensitive data on the control-plane bridge host.
• Use Case: An HPC operator can use this Skill to quickly build a personal or team read-only dashboard that shows running/pending jobs, per-account GPU-hour budget consumption, and SSD/inode storage health, all accessible via a public Vercel URL without exposing cluster credentials.

Quick Start

Use the sue-slurm-web-design skill to design, scaffold, and deploy a read-only Slurm job monitoring dashboard for your NM5 sandbox on Vercel, reusing Slurm-web's visual design and setting up the hourly SSH collector on your always-on control-plane host.