runpod

runpod is a skill for Claude Code from neuromechanist/research-skills. It costs 176 tokens per session (3,935 once invoked), scanned A, original, BSD-3-Clause.

A workflow for renting cloud GPUs through RunPod, a service that provides remote machines with graphics processors. It prepares the software in advance so a rented machine can start the workload quickly.

In plain words
What is it for?
It helps run workloads that need more memory, CUDA, or multiple GPUs than a local computer has. It covers GPU choice, image preparation, pod startup and shutdown, benchmarking, model training, fine-tuning, distillation, and short-lived model serving.
Why use it?
Installing packages and compiling software after a GPU machine starts can waste paid rental time. This workflow moves that setup into a reusable container image and helps avoid unsuitable or unnecessary GPU rentals.

Skill for Claude Code

Written for Claude Code: shipped in a Claude Code plugin. Also seen: positional $N argument.

Part of the ml-training plugin — 1 skill shipped together

Good fit It helps run workloads that need more memory, CUDA, or multiple GPUs than a local computer has. It covers GPU choice, image preparation, pod startup and shutdown, benchmarking, model training, fine-tuning, distillation, and short-lived model serving.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/neuromechanist/research-skills/runpod
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.

Any agent
npx skills add neuromechanist/research-skills --skill runpod
Clone the repo
git clone --depth 1 https://github.com/neuromechanist/research-skills

Made for: Claude Code.

Or install ml-training, the plugin that ships this one along with the rest of its 1 skill.

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 runpod

README.md
[![agentmods](https://agentmods.dev/badge/skills/neuromechanist/research-skills/runpod/github.svg)](https://agentmods.dev/skills/neuromechanist/research-skills/runpod)
Your own site
<a href="https://agentmods.dev/skills/neuromechanist/research-skills/runpod"><img src="https://agentmods.dev/badge/skills/neuromechanist/research-skills/runpod/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.

agentmods 80×15 button for runpod

Your own site · 80×15
<a href="https://agentmods.dev/skills/neuromechanist/research-skills/runpod"><img src="https://agentmods.dev/badge/skills/neuromechanist/research-skills/runpod.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 176 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 3,935 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 1 finding. A grade says what 26 rules found in the file — not that it is safe. Third-party audits
  • NVIDIA SkillSpector warn 7 Sept 2026
SkillSpector: 4 findings, up to medium

These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →

  • medium Data Exfiltration · line 44
    Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
    Fix: Verify the destination URL is trusted and necessary. Remove or replace with documented APIs. Ensure no secrets, tokens, or PII are transmitted.
  • medium Data Exfiltration · line 44
    Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
    Fix: Verify the destination URL is trusted and necessary. Remove or replace with documented APIs. Ensure no secrets, tokens, or PII are transmitted.
  • medium MCP Rug Pull · line 65
    Docker image references without a specific tag (:latest is implicit) or digest (@sha256:...) can be silently replaced by a malicious image.
    Fix: Pin the image: image:tag or image@sha256:abc123
  • medium MCP Rug Pull · line 66
    Docker image references without a specific tag (:latest is implicit) or digest (@sha256:...) can be silently replaced by a malicious image.
    Fix: Pin the image: image:tag or image@sha256:abc123
How audits are shown
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 $0.00176 $0.03935
Opus 5 $0.00088 $0.01968
Sonnet 5 $0.00035 $0.00787
Haiku 4.5 $0.00018 $0.00394

Measured 12d ago against content hash 85b6d5f1660c, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-12, from the pricing page.

Security

Grade A, and why

runpod scanned grade A with 1 finding 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 12d ago.

The scan reads SKILL.md. This mod also ships 4 executable files (templates/pod-down.sh, templates/pod-run.sh, templates/pod-up.sh, …), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

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.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

curl -s https://api.runpod.io/graphql \
plugins/ml-training/skills/runpod/SKILL.md · 155 lines

How it starts

The opening of the file, as written. The whole thing — 155 lines — stays where its author put it; the contents beside it link to each section on GitHub.

RunPod Fast Provisioning

Rent a GPU that is ready in seconds, not minutes. The whole method is one rule: everything installable goes into a prebaked container image, so the pod's only job at boot is to pull that image, start sshd, and run the workload. Pod-side apt install, pip install, and source compilation are all billed minutes that buy nothing.

When to Use

  • Work that is impossible or impractically slow on local hardware: needs more VRAM than the workstation has, needs CUDA specifically, needs N GPUs on one host.
  • Benchmark grids, fine-tuning and distillation runs, short-lived model serving.
  • Any repeat rental: the second pod is where a prebaked image pays for itself.

Do not rent for work that fits locally. Write the case for the pod first (what is locally impossible, what decides when it is done), and give the workload a hard stop before launching it.

The core rule: prebake, never install on the pod

Stage Prebaked image Stock image plus pod-side installs
Boot to verified environment, warm host (image cached) 16 s 8-10 min
Boot to verified environment, cold host (1.9 GB image pull) 64 s 8-10 min

Measured 2026-08-12 on RunPod secure cloud; "verified" means pod created, RUNNING, ssh accepted, and an on-pod check of GPU count, binary version, and the prebaked virtual environment all passed. Related measurements from the same session:

  • A 17 GB model file pulled from HuggingFace on the pod in 34.8 s (about 490 MB/s) with hf_transfer preinstalled and enabled. Datacenter network beats any local upload.
  • 2x A40 at $0.88/hr total served a 30B model layer-split across both cards (8.2 + 8.6 GiB resident) at 31 tok/s decode.

Always report the GPU type, GPU count, hourly price, and the date next to any number like these. Prices and stock move, and a throughput number without its machine spec is not a result.

Why RunPod over a managed serverless platform

Compared to managed alternatives such as Modal, RunPod runs roughly half the price for the same card, needs minimal prep work, exposes a usable REST and GraphQL API, and sits on a fast datacenter network. The trade-off is that the image, the lifecycle, and the cost discipline are yours to manage, which is exactly what the templates in this skill do.

Read the full file on GitHub · 155 lines

Files

What ships with it

9 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

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. 12d ago First seen · 155 lines · 176 tokens per session scan A 85b6d5f1660c

Subscribe to this mod's changes

runpod is a skill published in the GitHub repository neuromechanist/research-skills (45 stars, last pushed 10d ago), licensed BSD-3-Clause. It adds 176 tokens to every session and 3,935 once invoked, about $0.0009 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other skills, from other repositories