What is resource-aware-ml?

A set of rules for writing PyTorch machine-learning and GPU code. It covers reduced-precision model loading, suitable inference servers for production APIs, and choosing a usable device such as a GPU, Apple accelerator, or CPU.

Use it when loading large transformer models, serving language models through an API, or writing GPU and machine-learning inference code.

It helps avoid running out of GPU memory and avoids code that fails when a particular accelerator is unavailable.

resource-aware-ml — Cursor rule for Cursor by j4flmao/agent-skills

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.

Clone the repo

git clone --depth 1 https://github.com/j4flmao/agent-skills

Made for: Cursor.

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.

README.md

[![agentmods](https://agentmods.dev/badge/rules/j4flmao/agent-skills/resource-aware-ml/github.svg)](https://agentmods.dev/rules/j4flmao/agent-skills/resource-aware-ml)

Your own site

<a href="https://agentmods.dev/rules/j4flmao/agent-skills/resource-aware-ml"><img src="https://agentmods.dev/badge/rules/j4flmao/agent-skills/resource-aware-ml/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/rules/j4flmao/agent-skills/resource-aware-ml"><img src="https://agentmods.dev/badge/rules/j4flmao/agent-skills/resource-aware-ml.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 16 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 334 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.

Model	Per session	Once invoked
Fable 5.1	$0.00016	$0.00334
Opus 5.5	$0.00006	$0.00134
Sonnet 5.5	$0.00003	$0.00067
Haiku 4.5	$0.00002	$0.00033

Measured 1mo ago against content hash e337f5db05e0, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

resource-aware-ml 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 1mo 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.

.cursor/rules/resource-aware-ml.mdc · 28 lines

What it actually says

Resource-Aware Machine Learning Rules

Cursor/AI MUST follow these rules when writing code for PyTorch, ML Inference, or GPU computations.

1. Ban FP32 for Large Models

Rule: Never initialize or load large Transformer models in float32 (FP32) unless explicitly requested.
Why: It instantly causes VRAM exhaustion (OOM).

Action: Always default to torch.float16 or torch.bfloat16.

# BAD
model = AutoModelForCausalLM.from_pretrained("llama")
# GOOD
model = AutoModelForCausalLM.from_pretrained("llama", torch_dtype=torch.bfloat16)

2. Enforce Inference Engines for APIs

Rule: If asked to "create a production API to serve an LLM", do NOT write a Flask/FastAPI wrapper around transformers.generate().
Action: You must propose or implement an enterprise-grade inference engine (e.g., vLLM or TGI) that supports continuous batching and PagedAttention.

3. Explicit Device Management

Rule: Never hardcode .to('cuda:0'). Always use dynamic device selection and gracefully fallback to CPU/MPS if CUDA is unavailable.
Action: Use device = torch.device("cuda" if torch.cuda.is_available() else "cpu").

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.

1mo ago First seen · 28 lines · 16 tokens per session scan A e337f5db05e0

Subscribe to this mod's changes

resource-aware-ml is a cursor rule published in the GitHub repository j4flmao/agent-skills (33 stars, last pushed 10d ago), licensed MIT. It adds 16 tokens to every session and 334 once invoked, about $0.0001 per session on Opus 5.5. 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 2026-09-05.

Data and AI

resource-aware-ml

Getting it into your agent

Wrote this? Show the measurements

What it costs to keep this loaded

Grade A, and why

Nothing flagged

What it actually says

Resource-Aware Machine Learning Rules

1. Ban FP32 for Large Models

2. Enforce Inference Engines for APIs

3. Explicit Device Management

What this file has done since we first saw it

Other cursor rules, from other repositories

tensorflow

006_Program_of_Thought_Tutorial

standards-data-eng

ponytail

angular-20

dev-standard