Slurm Job Array Starter

SkillAI & models

Use this skill to submit a tiny real Slurm job array, wait for terminal accounting, and summarize per-task outputs in compact JSON.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the Slurm Job Array Starter skill

About this capability

A framework for discovering, compiling, and validating reusable skills for scientific agents.

What this skill tells your AI

The instructions your AI receives, as published by ma-compbio-lab/skillfoundry in skills/hpc/slurm-job-array-starter/SKILL.md and read by ahel’s review.

Use this skill to submit a tiny real Slurm job array, wait for terminal accounting, and summarize per-task outputs in compact JSON.

What it does

  • Renders and submits a minimal sbatch --array script on the live cluster.
  • Polls queue state with squeue and waits for terminal task accounting with sacct.
  • Reads each task's stdout and stderr logs and exports a per-task summary.

When to use it

  • You need a verified starter for the job arrays leaf.
  • You want a minimal example of per-task output capture and accounting on the current Slurm cluster.
  • You need deterministic JSON for tests before scaling to larger batched workloads.

Example

python3 skills/hpc/slurm-job-array-starter/scripts/run_slurm_job_array.py \
  --partition cpu \
  --array-spec 0-1 \
  --sleep 1 \
  --out scratch/slurm-array/summary.json

Verification

  • Skill-local tests: python3 -m unittest discover -s skills/hpc/slurm-job-array-starter/tests -p 'test_*.py'
  • Expected summary: task_count == 2 and all task states are COMPLETED with ExitCode 0:0

Signals

GitHub stars
39
Forks
5
Last commit
Sep 2026
Advanced
Catalog kind
skill
Gateway key
slurm-job-array-starter
Source
github.com/ma-compbio-lab/skillfoundry