Skip to main content
affpapa.org
RU EN
skill CX Codex

tilegym-adding-cutile-kernel — Claude Code skill: how to install

Guides AI coding agents through adding cuTile GPU kernel operators to TileGym. Useful when implementing custom kernels, tests, and benchmarks.

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓ 
npx skills add NVIDIA/skills --skill tilegym-adding-cutile-kernel -a claude-code
Open on GitHub

What tilegym-adding-cutile-kernel does

This skill defines an ordered workflow for implementing cuTile backend operators. The agent first tracks the six-step checklist in TodoWrite before editing code. Work starts in src/tilegym/ops/ops.py, defining the @dispatch interface that throws NotImplementedError for unhandled backends.

Next, the kernel is written with cuda.tile and registered via @register_impl. Without explicit exports in src/tilegym/ops/cutile/__init__.py, the cuTile implementation fails to load at runtime. The workflow finishes with PyTorch reference tests and Triton-based performance benchmarks measuring TFLOPS or bandwidth.

What it helps with

  • register dispatch interfaces in ops.py
  • implement GPU kernels using cuda.tile
  • export cuTile modules in __init__.py
  • write correctness tests against PyTorch references
  • build operator benchmarks using Triton testing tools

Example prompts for the agent

  • Add a new cuTile kernel operator named my_op to TileGym
  • Register a cuTile backend implementation and create benchmarks
  • Set up tests and dispatch interface for a new cuTile op

How to install tilegym-adding-cutile-kernel

1Skills CLI (Vercel)

npx skills add NVIDIA/skills --skill tilegym-adding-cutile-kernel -a claude-code

2Manually: put the files in place

~/.claude/skills/tilegym-adding-cutile-kernel/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/tilegym-adding-cutile-kernel/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about tilegym-adding-cutile-kernel

How to install tilegym-adding-cutile-kernel in Claude Code?

Run npx skills add NVIDIA/skills --skill tilegym-adding-cutile-kernel -a claude-code in your terminal.

What does tilegym-adding-cutile-kernel do?

It instructs coding agents on the exact sequence to register, implement, test, and benchmark cuTile kernels in TileGym.

Why is registration in __init__.py required?

The cuTile operator backend will never load unless it is explicitly imported and added to __all__ inside src/tilegym/ops/cutile/__init__.py.

Similar skills

Everything for AI and ML →
model-serving-kubernetes devops-skills Deploys ML models and LLMs on Kubernetes using KServe and NVIDIA Triton Inference Server. Handles canary rollouts, traffic splitting, and GPU scheduling. CX Codex +1 327installs runpod-templates runpod A skill that finds official Runpod Pod templates, explains their ports and autostart, and repairs model metadata in ComfyUI workflows. CX Codex +1 10,100installs deep-learning Mindrally Gives your agent a rulebook for neural network work: layer design, optimizers, mixed precision, memory. Useful when you write or refactor model code. CX Codex +1 946installs tracelens-analysis-orchestrator amd A skill that runs the agentic TraceLens workflow over PyTorch Profiler traces and writes a prioritized analysis.md. CX Codex +1 877installs quark-install amd Installs or verifies AMD Quark and makes sure the Python environment has a PyTorch build that matches the accelerator. Useful when the quark import fails or kernels won't compile. CX Codex +1 526installs vast-gpu wanshuiyin Evaluates training tasks, rents cost-effective GPU instances on vast.ai, and manages their lifecycle. Handy when you need cloud compute without local hardware. CX Codex +1 453installs

From the same set178

holoscan-install-wheel Installs Holoscan SDK 5.x CUDA 13 Python wheels into an isolated venv and checks that the import works. CX Codex +1 1,800installs holoscan-install-conda Checks Conda package availability for Holoscan SDK 5.x releases, flags missing 5.0 builds, and routes to supported installation alternatives. CX Codex +1 1,800installs holoscan-install-container Checks a host and installs the Holoscan SDK 5.x CUDA 13 development container. Use it before running the SDK in Docker with GPU access. CX Codex +1 1,800installs holoscan-setup Checks a Linux host and picks a Holoscan SDK 5.x install path: container, Debian packages, wheels or a source build. CX Codex +1 1,800installs mcore-testing A skill that maps out the Megatron-LM test system: test layout, recipe YAML, golden values, and how to run unit tests locally. CX Codex +1 1,800installs mcore-create-issue Investigates failing GitHub Actions runs for NVIDIA/Megatron-LM, extracts the error cause, and files a structured bug report. CX Codex +1 1,800installs mcore-run-on-slurm Prepare a SLURM launch for distributed Megatron-LM training, from an sbatch skeleton to rank-level failure checks. Useful when configuring or debugging a multi-node job. CX Codex +1 1,800installs nemo-rl-brev-etiquette A skill that tells the agent where files belong on a Brev box: code stays in the repo, heavy run output goes to /ephemeral. CX Codex +1 1,800installs

Project sponsors

Companies that keep this analytics open
Gold sponsor
Nutra.Club

Impressions, clicks and code copies — open statistics. We keep the numbers public for the same reason NeBlask itself is public. Become a sponsor