skills.rest
EmaRimoldi
EmaRimoldiCOMMUNITY
Explore all skills by this author →
GitHub

experiment-runner

Generates run manifests and launches SLURM jobs for phased HPC experiments.

Updated Mar 27, 2026
Search Tags:#workflow#monitoring#reproducibility#experiment#hpc#slurm#phase-gate
One-click install
npx skills add https://github.com/EmaRimoldi/Claude-scholar-extended --skill experiment-runner
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: experiment-runner
Source: https://github.com/EmaRimoldi/Claude-scholar-extended/tree/main/skills/experiment-runner
Command: npx skills add https://github.com/EmaRimoldi/Claude-scholar-extended --skill experiment-runner

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

The Experiment Runner coordinates end-to-end HPC experiments by translating an experiment design into a runnable SLURM-based matrix, handling phased gates, failure recovery, and reproducibility.

Core Features & Use Cases

  • Phase-based execution: constructs and submits run matrices, assigns deterministic run IDs, and tracks progress across phases.
  • Failure handling and resubmission: automatically detects failures, applies recovery policies, and resubmits when appropriate.
  • Reproducibility: captures seeds, environment, and configurations per run, and updates a centralized state file for traceability.

Quick Start

Provide your experiment-plan.md and SLURM scripts, then ask the skill to launch Phase 1 runs on your HPC cluster.

Frequently Asked Questions about experiment-runner

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run reproducible HPC experiments across multiple phases using SLURM?▼

To run reproducible HPC experiments, this skill translates your experiment design into a runnable SLURM-based matrix, handling phased gates, failure recovery, and centralized state tracking for full traceability across phases.

What is the best way to automatically resubmit failed SLURM jobs in an HPC experiment matrix?▼

Automatically resubmitting failed SLURM jobs is handled by built-in failure detection, which applies recovery policies and resubmits runs when appropriate, updating the experiment-state.json file with per-run outputs.

How do I track experiment progress and manage resource allocation across phased SLURM gates?▼

Tracking experiment progress across phased gates is achieved by generating a run manifest with deterministic run IDs, monitoring phase-gate decisions, and updating a centralized state file with resource allocation and checkpointing data.

Do I need a specific experiment plan format to launch phased SLURM runs?▼

You need to provide an experiment-plan.md file alongside your SLURM scripts to launch phased runs, which allows the system to construct and submit the run matrices on your HPC cluster.

How does seed management and environment capture work for HPC experiment reproducibility?▼

Seed management and environment capture for HPC reproducibility work by recording configurations, seeds, and environment details per run, ensuring every phase-gate execution can be traced and replicated exactly.

What are the limitations of using phase-gate execution for large-scale SLURM job monitoring?▼

Phase-gate execution for large-scale SLURM monitoring requires HPC cluster access and pre-defined SLURM scripts, meaning it is not suited for local or non-HPC experiment execution environments.

Related Skills

Explore More Education & Research Skills→
EmaRimoldiEmaRimoldi

compute-planner

Generates SLURM scripts and computes resource allocations from experiment matrices and budgets.

Community
Advanced
MohamedAbdallah-14MohamedAbdallah-14
3

wz:compliance-measurement

Measure pipeline compliance for completed Wazir runs and produce per-phase and aggregate scores.

Community
Advanced
michaelrizvimichaelrizvi
11

slurm

Write, debug, and optimize SLURM job scripts for HPC clusters.

Community
Intermediate
KangOxfordKangOxford

find-wandb

Locate WandB run URLs, directories, and configurations for SLURM jobs.

Community
Advanced