What is sbatch-scripts?

A set of rules for SLURM job-submission scripts used in model examples. SLURM is software that queues and runs jobs on shared computing clusters.

Use it when creating or editing sbatch scripts for AMD GPUs, especially scripts that run with Apptainer and need portable partition and account settings.

It prevents common cluster issues such as incorrect script paths, NVIDIA-only GPU options, hard-coded site settings, and unclear account placeholders.

sbatch-scripts — Cursor rule by AMDResearch

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods

npx agentmods add rules/amdresearch/ai4science-studio/sbatch-scripts

Clone the repo

git clone --depth 1 https://github.com/AMDResearch/ai4science-studio

Made for: Cursor.

Per session 0 Nothing until a file matches its globs; then the whole rule loads.

When invoked 827 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. Scan, not verified.

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.

Model	Per session	Once invoked
Fable 5	$0.00000	$0.00827
Opus 5	$0.00000	$0.00413
Sonnet 5	$0.00000	$0.00165
Haiku 4.5	$0.00000	$0.00083

Measured yesterday against content hash a0e580f8a3b2, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

sbatch-scripts 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 yesterday.

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.

.cursor/rules/sbatch-scripts.mdc · 66 lines

How it starts

The opening of the file, as written. The whole thing — 66 lines — stays where its author put it; the contents beside it link to each section on GitHub.

SLURM sbatch script conventions

When creating or editing sbatch_*.sh under any model's examples/ directory:

Required patterns

SCRIPT_DIR resolution — Use scontrol to recover the original script path (SLURM copies scripts to /var/spool/slurmd/):

if [[ -n "${SLURM_JOB_ID:-}" ]]; then
    SCRIPT_DIR=$(cd "$(dirname "$(scontrol show job "$SLURM_JOB_ID" | grep -oP 'Command=\K\S+')")" && pwd)
else
    SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
fi

Use --rocm for AMD GPU passthrough — Never use --nv (that is for NVIDIA):
```
apptainer exec --rocm ...
```
Naming convention — Scripts are named sbatch_<task>_amd.sh, not _mi300x.sh or _mi250x.sh. The same rocm7.2.x image covers MI250X (gfx90a), MI300X (gfx942), and MI350X (gfx950).
Placeholder partition/account — Use YOUR_PARTITION_HERE and YOUR_ACCOUNT_HERE in #SBATCH directives so users (or agents) know to replace them:
```
#SBATCH --partition=YOUR_PARTITION_HERE
#SBATCH --account=YOUR_ACCOUNT_HERE
```
No site-specific paths — including in defaults — Never hardcode user- or cluster-specific paths anywhere in the script, including as default values for env vars. This is the most common mistake: satisfying "paths come from env vars" while still baking a site-specific path into the default.

Wrong:
```
HG_SIF="${HG_SIF:-/shared/username/models/HydraGNN/images/rocm_pytorch.sif}"
```
Right — use AI4S_SHARED_DIR (set by the user for their site) or leave the default empty and fail with a clear message:
```
# Option A: relative to a user-set base dir
HG_BASE="${AI4S_SHARED_DIR:-}/models/HydraGNN"
HG_SIF="${HG_SIF:-${HG_BASE}/images/rocm_pytorch.sif}"

# Option B: no default — require the user to set it
: "${HG_SIF:?Set HG_SIF to your Apptainer SIF path}"
```
Site-specific values (partition, account, shared dir, scratch paths) belong in the gitignored .cluster-config.yaml / ~/.config/ai4science-studio/cluster.yaml, not in committed scripts.

Read the full file on GitHub · 66 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.

yesterday First seen · 66 lines · 0 tokens per session scan A a0e580f8a3b2

Subscribe to this mod's changes

sbatch-scripts is a cursor rule 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 827 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.

Data and AI Scientific ai-for-science amd drug-discovery earth-science fine-tuning healthcare

sbatch-scripts

Getting it into your agent

What it costs to keep this loaded

Grade A, and why

Nothing flagged

How it starts

SLURM sbatch script conventions

Required patterns

What this file has done since we first saw it

Other cursor rules, from other repositories

angular-20

dev-standard

typescript

coolify-ai-docs

python_lib

specs