Agent skill

pytorch-cuda

View SKILL.md on GitHub Repository
Stars 0
Forks 0

Install this agent skill to your Project

npx add-skill https://github.com/cuba6112/skillfactory/tree/main/skills/pytorch-cuda

SKILL.md

PyTorch CUDA

Overview

Configure and verify CUDA 13 readiness (toolkit, driver, and PyTorch wheel support), then run PyTorch CUDA with reliable timing and memory practices.

When to Use

Use this skill only when the frontmatter triggers apply; otherwise start with a standard PyTorch setup or CPU run.

Decision Tree

  1. Are you targeting CUDA 13?
    • Yes: verify driver minimums and toolkit availability.
  2. Do you rely on prebuilt PyTorch wheels?
    • Yes: confirm the available CUDA builds (11.8/12.6/12.8) before upgrading.
  3. Do you need GPU profiling accuracy?
    • Yes: use CUDA events and explicit synchronization.

Workflows

1. CUDA 13 Readiness Check

  1. Read the installed driver version (nvidia-smi).
  2. Compare against CUDA 13.0 minimums (GA/Update 1/Update 2).
  3. Confirm toolkit availability (nvcc --version).
  4. Probe PyTorch runtime (torch.version.cuda, torch.cuda.is_available()).

2. PyTorch Wheel Compatibility Gate

  1. Check the PyTorch install selector for CUDA wheel versions.
  2. If the required CUDA version is not listed, keep the current CUDA or build from source.
  3. Pin PyTorch wheel URLs to the listed CUDA build for reproducibility.

3. Accurate CUDA Timing

  1. Create torch.cuda.Event(enable_timing=True) start/end events.
  2. Record start, run the kernel, record end.
  3. Synchronize the end event before reading elapsed time.

Non-Obvious Insights

  • CUDA toolkit components are versioned independently, so library versions can diverge from the toolkit label.
  • CUDA 13 driver minimums differ across 13.0 GA and updates; verify the exact update target.
  • PyTorch prebuilt wheels list CUDA 11.8, 12.6, and 12.8; plan CUDA 13 adoption accordingly.

Evidence

  • "CUDA Toolkit 13.1.0 (December 2025), Versioned Online Documentation" - NVIDIA
  • "Starting with CUDA 11, individual components within the CUDA Toolkit (for example: compiler, libraries, tools) are versioned independently." - NVIDIA
  • "CUDA 13.0 Update 2 >=580.95.05 N/A CUDA 13.0 Update 1 >=580.82.07 N/A CUDA 13.0 GA >=580.65.06 N/A" - NVIDIA
  • "CUDA 11.8 CUDA 12.6 CUDA 12.8" - PyTorch

Scripts

  • scripts/pytorch-cuda_tool.py: CLI for probing CUDA versions and checking driver minimums.
  • scripts/pytorch-cuda_tool.js: Node.js CLI for the same checks.

Dependencies

  • Python 3.11+ or Node 18+.

References

  • references/README.md

Recommended Agent Skills

Expand your agent's capabilities with these related and highly-rated skills.

cuba6112/skillfactory

prompt-engineering

Comprehensive prompt engineering techniques for Claude models. Use this skill when crafting, optimizing, or debugging prompts for Claude API, Claude Code, or any Claude-powered application. Covers system prompts, role prompting, multishot examples, chain of thought, XML structuring, long context handling, extended thinking, prompt chaining, Claude 4.x-specific best practices, and agentic orchestration including subagents, agent loops, skills, MCP integration, and multi-agent workflows.

0 0
Explore
cuba6112/skillfactory

adk-rag-agent

Build RAG (Retrieval-Augmented Generation) agents with Google ADK and Vertex AI RAG Engine. Use when implementing document Q&A, knowledge base search, or citation-backed responses. Covers VertexAiRagRetrieval tool, corpus setup, and citation formatting.

0 0
Explore
cuba6112/skillfactory

headless-cli-agents

Build agentic systems using Claude CLI in headless mode or the Claude Agent SDK. Use when building automation pipelines, CI/CD integrations, multi-agent orchestration, or programmatic Claude interactions. Covers CLI flags (-p, --output-format), session management (--resume, --continue), Python SDK (claude-agent-sdk), custom tools, and agent loop patterns.

0 0
Explore
cuba6112/skillfactory

notion-knowledge-capture

Capture conversations and decisions into structured Notion pages; use when turning chats/notes into wiki entries, how-tos, decisions, or FAQs with proper linking.

0 0
Explore
cuba6112/skillfactory

mcp-builder

Guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK).

0 0
Explore
cuba6112/skillfactory

gh-fix-ci

Inspect GitHub PR checks with gh, pull failing GitHub Actions logs, summarize failure context, then create a fix plan and implement after user approval. Use when a user asks to debug or fix failing PR CI/CD checks on GitHub Actions and wants a plan + code changes; for external checks (e.g., Buildkite), only report the details URL and mark them out of scope.

0 0
Explore

Didn't find tool you were looking for?

Be as detailed as possible for better results