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.
npx agentmods add skills/tensormux/kernel-skills/write-kernel-test-plannpx skills add tensormux/kernel-skills --skill write-kernel-test-plangit clone --depth 1 https://github.com/tensormux/kernel-skillsWrote 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.
[](https://agentmods.dev/skills/tensormux/kernel-skills/write-kernel-test-plan)<a href="https://agentmods.dev/skills/tensormux/kernel-skills/write-kernel-test-plan"><img src="https://agentmods.dev/badge/skills/tensormux/kernel-skills/write-kernel-test-plan.svg" alt="Measured on agentmods" height="20"></a>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 | $0.00000 | $0.02775 |
| Opus 5 | $0.00000 | $0.01388 |
| Sonnet 5 | $0.00000 | $0.00555 |
| Haiku 4.5 | $0.00000 | $0.00278 |
Grade A, and why
write-kernel-test-plan 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 4d 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.
How it starts
The opening of the file, as written. The whole thing — 141 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Skill: Write a Kernel Test Plan
Purpose
Guide the agent through constructing a systematic, coverage-complete test plan for a compute kernel, covering correctness, numerical precision, boundary conditions, layout variations, and performance regression.
Use this when
- Writing a new CUDA or Triton kernel that needs a test suite before it is used in production.
- Reviewing an existing kernel where the test coverage is unknown or suspected to be incomplete.
- A kernel has exhibited silent correctness failures and the root cause needs to be isolated through systematic testing.
- Preparing a kernel for open source release or external contribution.
Do not use this when
- The kernel is a trivial wrapper around a well-tested library call (e.g., a single cuBLAS invocation) with no added logic.
- The "kernel" is a Python-level composition of existing tested ops with no custom GPU code.
Inputs the agent should gather first
- The mathematical specification of the kernel: what function does it compute, exactly?
- Input and output dtypes.
- The set of tensor shapes the kernel is expected to handle, including whether shapes are static or dynamic.
- Memory layout assumptions: does the kernel require contiguous input? Does it handle strided tensors?
- Whether the kernel has stochastic behavior (e.g., dropout) that requires special handling in tests.
- The target hardware: some kernels have architecture-specific code paths (e.g., tensor core paths vs fallback paths) that must be tested separately.
- Whether there is an existing reference implementation (PyTorch op, numpy, or a simpler known-correct version) to compare against.
Required reasoning process
- Define the reference implementation. Before writing any test, identify the ground truth:
- Prefer a CPU reference in fp64 or fp32 for maximum precision.
- For kernels that match an existing PyTorch op, use
torch.<op>as the reference. - For kernels with no direct PyTorch equivalent, write a simple, unoptimized reference in Python/numpy that is obviously correct.
- The reference must not share code with the kernel under test. It must be independently correct.
What ships with it
1 file 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.
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.
- 4d ago First seen · 141 lines · 0 tokens per session scan A 522b7196dbd0
write-kernel-test-plan is a skill published in the GitHub repository tensormux/kernel-skills (72 stars, last pushed 2mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 2,775 tokens. 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.
Other skills, from other repositories
design-impl-audit
The Design-vs-Implementation Audit Copilot. A universal skill for any project, any language, any design document. Before work starts, it reminds engineers what must be delivered; after code lands, it checks what was actually delivered, what drifted, and what silently went beyond the design. Feed it a design doc plus a…
skill-creator
Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.
diagnose
Disciplined diagnosis loop for hard bugs and performance regressions. Reproduce → minimise → hypothesise → instrument → fix → regression-test. Use when user says "diagnose this" / "debug this", reports a bug, says something is broken/throwing/failing, or describes a performance regression.
prompt-engineering
Prompt engineering techniques and patterns. Use when writing agent commands, hooks, skills, subagent prompts, or any LLM interaction: optimizing prompts, improving output reliability, and designing production-grade prompt templates. Trigger words: prompt engineering, prompt, prompt optimization, LLM interaction.
setup-matt-pocock-skills
Sets up an ## Agent skills block in AGENTS.md/CLAUDE.md and docs/agents/ so the engineering skills know this repo's issue tracker (GitHub or local markdown), triage label vocabulary, and domain doc layout. Run before first use of to-issues, to-prd, triage, diagnose, tdd, improve-codebase-architecture, or zoom-out — or…
brainstorming
Brainstorming design process. Must be used before any creative work - creating features, building components, adding functionality, or modifying behavior. Transforms ideas into complete designs and specifications through collaborative dialogue. Trigger words: brainstorm, design discussion, requirement analysis…