SKILLEMALL.ai
GitHub ↗ RU Sign in

BC kernel-triton-writing

ONLY for OpenAI Triton (@triton.jit) kernel development. NEVER use for CUDA C++ kernels, CuTe DSL kernels, TileIR, or profiling tools (ncu, nsys). The user's request must involve Triton explicitly. Covers Triton-specific patterns: fused elementwise, reductions (softmax, LayerNorm, RMSNorm), tiled GEMM with triton.autotune, and flash attention. Workflow: design, write, verify (with fast-path for explicit requests).

The skillemall take

A skill for developing GPU kernels with OpenAI Triton, the framework that lets you write GPU code without CUDA C++. Promises help with fused operations, reductions (softmax, LayerNorm), GEMM with autotune, and flash attention.

The package holds 13 files and 3 scripts; description spans 3665 tokens. Safety scores 95, quality 90, process 59. One medium-level finding, nothing critical. Code passed linting. No model runs or sandbox tests—actual performance is untested.

Install it if you're writing Triton kernels and need patterns for reductions and GEMM. If you work with CUDA C++ or profiling, the skill will refuse by design and waste your time.

NVIDIA/TensorRT-LLM Agent Skills author: NVIDIA NOASSERTION 13 files · 3 scripts body ≈ 3 665 tokens Open the sourcegithub.com↗ analyzed 5 d ago

ONLY for OpenAI Triton (@triton.jit) kernel development. NEVER use for CUDA C++ kernels, CuTe DSL kernels, TileIR, or profiling tools (ncu, nsys). The user's…

As a process C 59/100 · Has gaps — weak spots: result and completion, inputs and preconditions, running it twice

GeneratorSoftware developmentAI and agentstype and topics are labelled automatically from the skill text
Runs in: Claude / Claude Code ChatGPT / Codex Gemini CLI Cursor GitHub Copilot / VS Code Qwen Code Kimi Code CLI Z.ai (GLM) DeepSeek (Deep Code) Mistral Vibe LangChain (deepagents) OpenClaw Goose OpenCode
JSON
Technical rating
B
93/100
safety, quality, tests
Safety 60%
95
Quality 40%
90
Run on models
none yet
Process rating
C
59/100
Has gaps
Result and completion w 14
0
Inputs and preconditions w 11
0
Running it twice w 4
30
the three weakest of ten parameters · all ten

What is at stake

Medium-severity findings: the skill is probably honest, but read what alarmed the scanner.

Dangerous commands medium severity

Below is the worst case for this category. The finding here is medium: the guard saw a sign, not a proof.

If you install

The skill contains commands that delete files, rewrite disks or execute code fetched from the network. The agent may run them without asking if it believes the instructions require it.

For the author

Replace destructive commands with safe equivalents that ask for confirmation, scope them to one folder, and stop piping curl into a shell: pin a version and a checksum.

How to improve

    For the model run — optional
    • Your own cases (evals/evals.json, 4–6 real requests with expected answers): the full check would then run those instead of a model-drafted suite.
    • A spec.yaml with trigger phrases and assertions — a behaviour contract for CI; `skilltest init` writes a template.

    Guard findings · 1

    ✓ No critical or high findings

    Medium and low: 1
    • medium Dangerous commands cmd-autorun-instruction SKILL.md:25
      Instructs the agent to auto-run a script on every session
      4. Always run `scripts/verify_kernel.py` to validate against the reference.

    Files scanned: 13. Evidence is masked. Grey chips explain why severity was lowered.

    Against the Agent Skills spec

    ✓ No remarks against the Agent Skills spec

    Process rating: all ten parameters 59/100

    • 0Result and completion. Does not say what the result is
    • 0Inputs and preconditions. Does not say what the process needs to start
    • 30Running it twice. 4 mutating operations with no state check
    • 50When it triggers. No condition that starts the skill
    • 60Tools and files. Uses tools (bash, python) that frontmatter does not declare
    • 100Steps. 28 steps
    • 100Failures and branches. 3 branches, has a failure section
    • 100Consistency. Name and required fields are in place
    • 100Execution cost. Instruction body is 3665 tokens
    • 100Progress reporting. Reports progress
    • medium Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing

    Everything here is measured from the skill text rather than judged by a model, so the numbers are checkable. A parameter weighs more when it is a more common reason for the process to stall.

    Quality signals

    • +5Description has no quoted example phrases that should trigger the skill
    • +3Output format is not stated: the model decides each time
    • -31 of 3 scripts are never mentioned in SKILL.md
    • +2Single-language instructions
    • +4Description says when NOT to use the skill
    • +3Description length 417: enough signal without eating the budget
    • +4Structure: 18 headings
    • +3Step-by-step instructions: 28 items
    • +4Has examples (9 code blocks)
    • +4Reference files are cited in the instructions (8 of 9)
    • +1License stated

    Quality base 70; lint remarks subtract, signals add up to 100. Result: 90.