cutlass-cpp-kernel

A guide for writing and working with CUTLASS and CuTe C++ GPU kernels. CUTLASS is NVIDIA's library of reusable building blocks for fast GPU computations, and CuTe provides tools for describing their layouts and data movement.

In plain words
What is it for?
Writing, debugging, porting, reviewing, and optimizing GEMM, pipeline, epilogue, grouped, sparse, FP8, FP4, or mixture-of-experts kernels.
Why use it?
It helps make sense of complex GPU template code, configuration choices, layout errors, and performance problems on Hopper or Blackwell GPUs.

Skill for Claude CodeCodex

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 skills/vipshop/cache-dit/cutlass-cpp-kernel
Any agent
npx skills add vipshop/cache-dit --skill cutlass-cpp-kernel
Clone the repo
git clone --depth 1 https://github.com/vipshop/cache-dit

Made for: Claude Code, Codex.

Per session 77 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,954 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 $0.00077 $0.01954
Opus 5 $0.00039 $0.00977
Sonnet 5 $0.00015 $0.00391
Haiku 4.5 $0.00008 $0.00195

Measured 2d ago against content hash 354cc3bd1933, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

cutlass-cpp-kernel 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 2d 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.

.copilot/skills/cutlass-cpp-kernel/SKILL.md · 174 lines

How it starts

The opening of the file, as written. The whole thing — 174 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CUTLASS and CuTe C++ Kernel Development

Goal

Use the workspace CUTLASS checkout to understand, implement, and optimize CUTLASS- or CuTe-based C++ kernels while keeping CuTe DSL Python authoring in a separate dedicated workflow.

When to Use

Use this skill when you need to:

  • study or modify CUTLASS GEMM, epilogue, pipeline, or collective-builder code
  • navigate CuTe C++ headers such as layout.hpp, tensor.hpp, swizzle.hpp, and atom/*
  • inspect CUTLASS example kernels for Hopper, Blackwell, FP8, FP4, grouped GEMM, sparse GEMM, or MoE patterns
  • reason about template choices such as tile shapes, kernel schedules, epilogue schedules, stage count, and copy or MMA atoms
  • debug CUTLASS template errors, layout mismatches, or schedule/configuration issues
  • review a rewrite plan from an existing C++ kernel to another CUTLASS- or CuTe-based implementation

Do not use this skill for:

  • CuTe DSL Python kernel authoring as the primary task; use cute-dsl-kernel
  • generic CUDA/PTX debugging with no CUTLASS or CuTe angle; use cuda-cpp-kernel
  • cache-dit integration plumbing by itself; pair with operator-migration

Scope Split

This skill is the main home for:

  • CUTLASS C++ source navigation
  • CuTe C++ header semantics
  • collective and pipeline configuration
  • example-based C++ implementation patterns

For CuTe DSL Python kernels, JIT flows, and generated CuTe DSL API reference files, switch to cute-dsl-kernel.

Workspace Source Location

In this workspace, the CUTLASS checkout is:

  • workspace-relative: vipshop/cutlass
  • shell path: /workspace/dev/vipshop/cutlass

Do not rely on skill-local repository mirrors, update scripts, or any agent-local install path.

Reference Style Rule

When citing CUTLASS sources, prefer workspace-relative paths such as:

  • vipshop/cutlass/include/cutlass/gemm/collective/
  • vipshop/cutlass/include/cute/layout.hpp
  • vipshop/cutlass/examples/49_hopper_gemm_with_collective_builder/

Use absolute shell paths only inside literal command examples when required.

Read the full file on GitHub · 174 lines

Files

What ships with it

7 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. 2d ago First seen · 174 lines · 77 tokens per session scan A 354cc3bd1933

Subscribe to this mod's changes

cutlass-cpp-kernel is a skill published in the GitHub repository vipshop/cache-dit (1,262 stars, last pushed 4d ago), licensed Apache-2.0. It adds 77 tokens to every session and 1,954 once invoked, about $0.0004 per session on Opus 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 2026-08-30.

Related

Other skills, from other repositories

systematic-debugging

Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.

obra/superpowers · 21 tokens

next-cache-components-adoption

Turn on Cache Components in a Next.js app and resolve the blocking routes it surfaces. Use when the user wants to enable, adopt, or migrate to Cache Components, flip the cacheComponents flag, work through a flood of blocking-prerender / instant validation errors, run the cache-components-instant-false codemod, or…

vercel/next.js · 95 tokens

babysit-pr

Babysit a GitHub pull request after creation by continuously polling review comments, CI checks/workflow runs, and mergeability state until the PR is merged/closed or user help is required. Diagnose failures, retry likely flaky failures up to 3 times, auto-fix/push branch-related issues when appropriate, and keep…

openai/codex · 114 tokens

imagegen

Generate or edit raster images when the task benefits from AI-created bitmap visuals such as photos, illustrations, textures, sprites, mockups, or transparent-background cutouts. Use when Codex should create a brand-new image, transform an existing image, or derive visual variants from references, and the output…

openai/codex · 113 tokens

cpu-profile-analysis

Analyze V8/Chrome CPU profiles (.cpuprofile) and DevTools trace files (Trace-.json). Use when: profiling performance, investigating slow functions, comparing code paths, finding bottlenecks, analyzing timeToRequest, understanding call trees from sampling profiler data, analyzing layout/paint/rendering, investigating…

microsoft/vscode · 71 tokens

next-cache-components-optimizer

Drive a Next.js route to instant navigation by setting up an agentic loop, under Cache Components / PPR, on initial load (hard navigation) and client-side navigation (soft navigation). Encode the goal as a failing @next/playwright instant() e2e and work it to green, one verified route at a time; the shipped test then…

vercel/next.js · 170 tokens