analysis-to-slurm-runner

name	analysis_to_slurm_runner
description	Turn an analysis request into a Slurm-ready execution plan with commands, resource assumptions, and job structure.
category	bio/compute
version	1
requires_tools	["search_knowledge_base","python_repl","write_file"]
requires_network	false
user_invocable	true
tags	["slurm","pipeline","execution","hpc","runbook"]
aliases	["slurm_analysis_planner"]
species	any
modality	compute
stage	reporting
stability	evolving
safety_level	medium

Analysis to Slurm Runner

Purpose

Translate a biological analysis request into a cluster-oriented run plan, including Slurm assumptions, commands, and a draft script when appropriate.

When to use

Use this skill when the user wants to move from analysis planning to execution on an HPC cluster.

Required inputs

• analysis goal: what should be run
• toolchain (optional): python, R, scanpy, nextflow, custom script
• resource hints (optional): GPU, memory, CPUs, wall time
• output path (optional): where a script should be saved

Steps

1. Search local compute or pipeline guidance with search_knowledge_base.
2. Infer the likely resource profile from the requested analysis.
3. Use python_repl to organize the run plan if needed.
4. If a script is requested, draft a Slurm script or execution note using write_file.
5. Keep the output at the draft-plan level unless the user explicitly asks for a runnable script path.

Output format

• Run plan
• Resource assumptions
• Suggested command structure
• Optional script path

Failure modes

• Missing execution details: provide a conservative template and list assumptions.
• Unknown pipeline: ask the user what script or entrypoint should be run.
• Unsafe submission request: do not submit without the user's clear instruction.

Examples

• "Turn this scanpy workflow into a Slurm run plan."
• "Prepare a cluster script for a perturb-seq analysis job with 16 CPUs and 128G memory."