preflight-scripts

A set of rules for Python preflight checks that verify whether the required software and AMD GPU operations work before running a model.

In plain words
What is it for?
Use it to create preflight_slug.py checks for PyTorch, GPU access, tensor operations, model imports, and optional dependencies.
Why use it?
It catches missing packages, unavailable GPUs, and failed GPU calculations early, with clear pass or fail output and an exit code.

Cursor rule for Cursor

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/preflight-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 330 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.

ModelPer sessionOnce invoked
Fable 5 $0.00000 $0.00330
Opus 5 $0.00000 $0.00165
Sonnet 5 $0.00000 $0.00066
Haiku 4.5 $0.00000 $0.00033

Measured yesterday against content hash d65524c70144, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

preflight-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/preflight-scripts.mdc · 45 lines

What it actually says

Preflight script conventions

When creating or editing preflight_<slug>.py under any model's examples/ directory:

Required pattern

Use the standard check() helper with colored PASS/FAIL output:

PASS = "\033[32mPASS\033[0m"
FAIL = "\033[31mFAIL\033[0m"
errors = 0

def check(label, fn):
    global errors
    try:
        result = fn()
        print(f"[{PASS}] {label}" + (f": {result}" if result else ""))
    except Exception as exc:
        print(f"[{FAIL}] {label}: {exc}")
        errors += 1

Required checks (in order)

  1. import torch — Report version
  2. GPU accessible — torch.cuda.is_available() must be True; report device name
  3. GPU tensor ops — Create a tensor on GPU, do a matmul, report result
  4. Model-specific imports — Import the key packages the model needs
  5. Optional deps — Use try/except and report "not installed (optional)" rather than failing

Exit code

Exit with code 1 if any required check failed; 0 if all passed. Print a summary line at the end.

Naming

File: preflight_<slug>.py where <slug> matches the model folder name (lowercase, hyphens to underscores).

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. yesterday First seen · 45 lines · 0 tokens per session scan A d65524c70144

Subscribe to this mod's changes

preflight-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 330 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.