ai4science-studio: Command for Claude Code

.claude/commands/run-neuralgcm.md

run-neuralgcm is a command for Claude Code from AMDResearch/ai4science-studio. It costs 0 tokens per session (1,016 once invoked), scanned A, original, MIT.

A guided command for running NeuralGCM, a weather and climate prediction model, on an AMD computing cluster. It supports Docker containers or Apptainer on HPC systems and can use SLURM, a cluster job scheduler.

In plain words
What is it for?
Use it to configure and launch NeuralGCM inference, including checkpoint selection, container setup, data preparation, and cluster execution.
Why use it?
It gathers the required runtime, checkpoint, data, and cluster choices before starting, reducing setup mistakes during a GPU run.

Command for Claude Code

Written for Claude Code: $ARGUMENTS substitution.

This is AMDResearch/ai4science-studio's own configuration. It tells Claude Code how to work on ai4science-studio itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything ai4science-studio configures →

Reuse

Borrowing it

Nothing to install: this file belongs to AMDResearch/ai4science-studio. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/AMDResearch/ai4science-studio/main/.claude/commands/run-neuralgcm.md
Clone the repo
git clone --depth 1 https://github.com/AMDResearch/ai4science-studio

Made for: Claude Code.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for run-neuralgcm

README.md
[![agentmods](https://agentmods.dev/badge/commands/amdresearch/ai4science-studio/run-neuralgcm.svg)](https://agentmods.dev/commands/amdresearch/ai4science-studio/run-neuralgcm)
Your own site
<a href="https://agentmods.dev/commands/amdresearch/ai4science-studio/run-neuralgcm"><img src="https://agentmods.dev/badge/commands/amdresearch/ai4science-studio/run-neuralgcm.svg" alt="Measured on agentmods" height="20"></a>
Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,016 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00000 $0.01016
Opus 5 $0.00000 $0.00508
Sonnet 5 $0.00000 $0.00203
Haiku 4.5 $0.00000 $0.00102

Measured 8d ago against content hash b08d28344a41, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-08, from the pricing page.

Security

Grade A, and why

run-neuralgcm scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 8d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.claude/commands/run-neuralgcm.md · 110 lines

How it starts

The opening of the file, as written. The whole thing — 110 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Run NeuralGCM inference on an AMD cluster

Guide the user through running NeuralGCM end-to-end on an AMD cluster via Docker or SLURM.

Step 0 — Cluster config check

Check if .cluster-config.yaml (repo root) or ~/.config/ai4science-studio/cluster.yaml exists. If neither exists, run the /init-cluster flow first. If a config exists, read it and pre-fill container runtime and SLURM partition/account from saved values.

Step 1 — Questionnaire (ask ALL questions before doing anything)

Ask the user the following questions. Do not assume any defaults. Wait for answers to all questions before proceeding.

Q0. Container runtime Which container runtime?

  • Docker (recommended — docker_run.sh handles image + deps automatically)
  • Apptainer (for HPC — set NGC_SIF to SIF path built from rocm/dev-ubuntu-22.04:7.0.2-complete)

Q1. Checkpoint Which NeuralGCM checkpoint?

  • v1/deterministic_0_7_deg.pkl — 0.7° (~78 km, slowest, sharpest)
  • v1/deterministic_1_4_deg.pkl — 1.4° (~156 km, default)
  • v1/deterministic_2_8_deg.pkl — 2.8° (~312 km, fastest)
  • v1/stochastic_1_4_deg.pkl — stochastic 1.4°
  • v1_precip/stochastic_precip_2_8_deg.pkl — precipitation
  • v1_precip/stochastic_evap_2_8_deg.pkl — evaporation

Q2. Initial condition date What date for ERA5 initial conditions? (YYYY-MM-DD, default: 2020-01-01)

Q3. Forecast steps How many 6h forecast steps? (16 = 4 days, 40 = 10 days)

Q4. (Stochastic only) Seed JAX PRNG seed for stochastic checkpoints? (default: 0)

Q5. Partition and account (if SLURM) How should I determine your SLURM partition and account/project?

  • Provide manually — type your partition and account names
  • Auto-discover — I will query SLURM to find available partitions and accounts on this cluster

Step 2 — Act on answers

Auto-discovery procedures

Run these when the user chose Auto-discover for any question. Present the results and let the user confirm or override.

SLURM partition and account (Q5):

sinfo -h -o "%P %G" | grep -i gpu
sacctmgr show associations where user=$USER format=account%30,partition%30 -n

Present the available GPU partitions and the user's associated accounts. If multiple exist, ask the user to pick.

Read the full file on GitHub · 110 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 8d ago First seen · 110 lines · 0 tokens per session scan A b08d28344a41

Subscribe to this mod's changes

run-neuralgcm is a command published in the GitHub repository AMDResearch/ai4science-studio (4 stars, last pushed 1mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,016 tokens. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.