HPC tools

Open-source tools for shared Slurm clusters, from finding a free node to finding out why a job failed.

tools: 8
releases on PyPI: 86
downloads in the last 30 days: 2,071

A job's life, and the tool for each step

1
Plan

find room, write the script
- nodetop
- slurmate
2
Run

watch it as it works
- slurmwatch
3
Review

learn why it ended as it did
- slurmpast
4
Resize

ask for what it used
- cloudfit

and round again, asking for what the last run used

Storage

where the data lives, and how full it is

rapidu
dirscape

Assistant

the same questions, in plain English

hpcpilot

The tools

nodetop

Plan

What's actually free on your cluster, and why your job won't run there.

 87 partitions  · ❯8 open to you  ·  76 no access  ·  3 down     partition  nodes idle     mem free  cores free  gpus free  gpu model  usable    gpu-a            0/44  1320/11000G    396/1408     44/176  A100       ██▍░░░░░░░░░░░░░░░░░    compute          0/40  1400/10000G   1320/5120          -             ██▊░░░░░░░░░░░░░░░░░    wide            0/190   190/34200G    380/9120          -             ▏░░░░░░░░░░░░░░░░░░░

$ pip install nodetop

Version: 0.7.0
Releases: 9 since Aug 2026
Last 30 days: 446 downloads

Weekly downloads, last 13 weeks; the busiest week had 406

GitHub PyPI

slurmate

Plan

Stop hand-writing sbatch scripts. Let the wizard do it.

$ pipx install slurmate

Version: 0.8.0
Releases: 14 since Jun 2026
Last 30 days: 305 downloads

Weekly downloads, last 13 weeks; the busiest week had 314

GitHub PyPI

slurmwatch

Run

Is your Slurm job actually using what you asked for? Watch it live.

$ pip install slurmwatch

Version: 1.2.4
Releases: 36 since Jul 2026
Last 30 days: 344 downloads

Weekly downloads, last 13 weeks; the busiest week had 907

GitHub PyPI

Article slurmwatch: Live Telemetry for HPC Jobs

Article Correcting Claude Code's Cluster Resource Requests with slurmwatch: Requested Memory and CPU Cores Against Actual Usage

slurmpast

Review

Why your Slurm jobs failed, and what to ask for next time.

$ pip install slurmpast

Version: 0.9.0
Releases: 11 since Jul 2026
Last 30 days: 325 downloads

Weekly downloads, last 13 weeks; the busiest week had 473

GitHub PyPI

Article slurmpast: Post-Mortem Analysis for Finished HPC Jobs

cloudfit

Resize, Claude Code plugin

Fits Slurm jobs to what they actually used.

              allocated       used          cloudfit says cores        16  ██████████  5.4 ███▍      8    █████      ↓ cut RAM          96G ██████████  9.3G █        14G  █▌         ↓ cut walltime     2h  ██████████  47m ███▉      59m  ████▉      ↓ cut GPU HBM      80G ██████████  32G ████      72G  █████████  ↑ raise batch / seq / KV

$ claude plugin install cloudfit@cloudfit-plugin

GitHub