What problem does it solve? Setting up a fresh account on the University of Michigan Great Lakes HPC cluster involves dozens of cluster-specific quirks — LDAP-managed login shells, a 4 GB login-node memory cap, broken stow options, and no crontab — that turn onboarding into hours of trial and error. This Skill walks an agent through the entire setup in the correct order, explaining each step in plain language for users new to HPC. ## Core Features & Use Cases - Phased dotfiles installation: Clones the dotfiles repo and links packages with GNU Stow using the exact Great Lakes package list, avoiding other machines' configs. - Cluster-aware environment setup: Installs miniconda, micromamba, the py_general conda environment (on SLURM compute nodes to dodge the memory cap), Quarto, fzf, and vim plugins. - Documented quirks and fixes: Covers the broken stow --restow, the readonly TMOUT shell crash, old system tmux, compute-node proxy access, and SLURM account/partition conventions. - Use Case: A new lab member gets a Great Lakes account and asks the agent to set it up; the agent runs each phase in order, explains why bash must stay the login shell, and creates the analysis environment via an srun job. ## Quick Start Set up my new Great Lakes HPC account using the greatlakes-hpc-setup skill, starting with cloning the dotfiles and stowing the packages.
