Agent skill

Triton Skill

Write, debug, and optimize Triton and Gluon GPU kernels from local upstream tutorials, production kernels, language definitions, and compiler source.

MITAuto-check passedAI & LLM Engineering

Install Triton Skill

skills CLI

$ npx skills add slowlyC/agent-gpu-skills --skill triton-skill -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI

$ gh skill install slowlyC/agent-gpu-skills triton-skill --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy

$ git clone --depth 1 https://github.com/slowlyC/agent-gpu-skills.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/triton-skill .claude/skills/triton-skill && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Paste into Claude Code

Install the "triton-skill" agent skill from https://github.com/slowlyC/agent-gpu-skills/tree/main/skills/triton-skill into .claude/skills/triton-skill/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "triton-skill", then confirm the skill loads.

Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

$skill-installer

$skill-installer install https://github.com/slowlyC/agent-gpu-skills/tree/main/skills/triton-skill

Type this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.

skills CLI

$ npx skills add slowlyC/agent-gpu-skills --skill triton-skill -a codex

Project install goes to .agents/skills/; add -g for ~/.codex/skills/.

GitHub CLI

$ gh skill install slowlyC/agent-gpu-skills triton-skill --agent codex

Project scope by default (.agents/skills/); add --scope user for a personal install.

Manual copy

$ git clone --depth 1 https://github.com/slowlyC/agent-gpu-skills.git skills-src && mkdir -p .agents/skills && cp -r skills-src/skills/triton-skill .agents/skills/triton-skill && rm -rf skills-src

Use ~/.agents/skills/ instead of .agents/skills for a personal install.

Codex skills documentation · loads skills from .agents/skills/

Paste into Codex

Install the "triton-skill" agent skill from https://github.com/slowlyC/agent-gpu-skills/tree/main/skills/triton-skill into .agents/skills/triton-skill/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "triton-skill", then confirm the skill loads.

Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

skills CLI

$ npx skills add slowlyC/agent-gpu-skills --skill triton-skill -a cursor

Project install goes to .agents/skills/; add -g for ~/.cursor/skills/.

GitHub CLI

$ gh skill install slowlyC/agent-gpu-skills triton-skill --agent cursor

Project scope by default (.agents/skills/); add --scope user for a personal install.

Manual copy

$ git clone --depth 1 https://github.com/slowlyC/agent-gpu-skills.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/skills/triton-skill .cursor/skills/triton-skill && rm -rf skills-src

Use ~/.cursor/skills/ instead of .cursor/skills for a personal install.

Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/

Paste into Cursor

Install the "triton-skill" agent skill from https://github.com/slowlyC/agent-gpu-skills/tree/main/skills/triton-skill into .cursor/skills/triton-skill/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "triton-skill", then confirm the skill loads.

Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

gemini skills install

$ gemini skills install https://github.com/slowlyC/agent-gpu-skills.git --path skills/triton-skill

--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.

skills CLI

$ npx skills add slowlyC/agent-gpu-skills --skill triton-skill -a gemini-cli

Project install goes to .agents/skills/; add -g for ~/.gemini/skills/.

GitHub CLI

$ gh skill install slowlyC/agent-gpu-skills triton-skill --agent gemini-cli

Project scope by default (.agents/skills/); add --scope user for a personal install.

Manual copy

$ git clone --depth 1 https://github.com/slowlyC/agent-gpu-skills.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/skills/triton-skill .gemini/skills/triton-skill && rm -rf skills-src

Use ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.

Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/

Paste into Gemini CLI

Install the "triton-skill" agent skill from https://github.com/slowlyC/agent-gpu-skills/tree/main/skills/triton-skill into .gemini/skills/triton-skill/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "triton-skill", then confirm the skill loads.

Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

GitHub CLI

$ gh skill install slowlyC/agent-gpu-skills triton-skill

Installs for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).

skills CLI

$ npx skills add slowlyC/agent-gpu-skills --skill triton-skill -a github-copilot

Project install goes to .agents/skills/; add -g for ~/.copilot/skills/.

Manual copy

$ git clone --depth 1 https://github.com/slowlyC/agent-gpu-skills.git skills-src && mkdir -p .github/skills && cp -r skills-src/skills/triton-skill .github/skills/triton-skill && rm -rf skills-src

Use ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.

GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/

Paste into GitHub Copilot

Install the "triton-skill" agent skill from https://github.com/slowlyC/agent-gpu-skills/tree/main/skills/triton-skill into .github/skills/triton-skill/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "triton-skill", then confirm the skill loads.

GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

skills CLI

$ npx skills add slowlyC/agent-gpu-skills --skill triton-skill -a opencode

OpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.

GitHub CLI

$ gh skill install slowlyC/agent-gpu-skills triton-skill --agent opencode

Project scope by default (.agents/skills/); add --scope user for a personal install.

Manual copy

$ git clone --depth 1 https://github.com/slowlyC/agent-gpu-skills.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/skills/triton-skill .opencode/skills/triton-skill && rm -rf skills-src

Use ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.

OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/

Paste into OpenCode

Install the "triton-skill" agent skill from https://github.com/slowlyC/agent-gpu-skills/tree/main/skills/triton-skill into .opencode/skills/triton-skill/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "triton-skill", then confirm the skill loads.

OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

Facts

Skill name: triton-skill
GitHub stars: 169
Token cost: ~1.3k tokens
SKILL.md length: 413 words
Files: 2
Skills in repo: 4
Repo updated: Aug 8, 2026
First seen: Oct 7, 2026
Licence: MIT

At a glance

Write, debug, and optimize Triton and Gluon GPU kernels from local upstream tutorials, production kernels, language definitions, and compiler source.

Works in 5 steps: Identify whether the task is Triton,… → Find the closest current tutorial or… → Verify every API used against its… → …
The task explicitly involves triton.jit
SKILL.md covers Locate the checkout, Choose the source surface, Query workflow and Implementation discipline, plus 1 more section
Calls rg, bash and python3

What it does

Triton Skill is an agent skill from slowlyC/agent-gpu-skills. Write, debug, and optimize Triton and Gluon GPU kernels from local upstream tutorials, production kernels, language definitions, and compiler source. Use when the task explicitly involves triton.jit, triton.language, tl., Gluon, TensorDescriptor, Triton autotune, TritonGPU/MLIR lowering, tritonkernels, or converting a CUDA kernel to Triton. Use cuda-skill for raw CUDA/PTX and NVIDIA architecture facts, and cutlass-skill for CUTLASS, CuTe, or CuTeDSL work.

Its SKILL.md is about 1.3k tokens, which your agent loads only when the skill is triggered. The skill folder holds 1 other file (for example `quick-reference.md`).

It sits in AI & LLM Engineering, covering GPU and accelerator computing. It works with CUDA, NVIDIA AI Platform and Python. The licence is MIT.

When your agent uses it

The task explicitly involves triton.jit
Triton.language
TensorDescriptor
Triton autotune

Example prompts

“/triton-skill”

Requirements

Python 3

Workflow steps

5 steps, taken from the first numbered list in SKILL.md.

Identify whether the task is Triton, Gluon, or compiler internals.
Find the closest current tutorial or implementation for the operation and architecture.
Verify every API used against its definition or another current call site.
Preserve the target workload's shape, dtype, stride, layout, masking, and numerical contract.
Establish correctness before changing launch parameters or optimization strategy.

What it can do on your machine

Read from SKILL.md and the folder at commit ae02d07. It shows what the files ask for, not the result of running them.

Tool permissions
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Runs code
Shell commands in SKILL.md call:
- rg
- bash
- python3
From the folder's file list and the shell code blocks in SKILL.md.
Network
No URLs in SKILL.md.
From URLs in SKILL.md, links to its own repository left out.
Credentials
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Triton Skill loads about 1.3k tokens when it runs. Until then it costs about 119 tokens; SKILL.md has 413 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it: ~119
When it runs · the whole SKILL.md, loaded when a task matches: ~1.3k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from slowlyC/agent-gpu-skills at commit ae02d07, republished under its MIT licence (© slowlyC). 413 words, ~1,331 tokens.

Download SKILL.mdSave it as .claude/skills/triton-skill/SKILL.md (or your agent's skills folder). This skill also uses 1 other file; get the full folder from GitHub.

name: triton-skill
description: Write, debug, and optimize Triton and Gluon GPU kernels from local upstream tutorials, production kernels, language definitions, and compiler source. Use when the task explicitly involves triton.jit, triton.language, tl.*, Gluon, TensorDescriptor, Triton autotune, TritonGPU/MLIR lowering, triton_kernels, or converting a CUDA kernel to Triton. Use cuda-skill for raw CUDA/PTX and NVIDIA architecture facts, and cutlass-skill for CUTLASS, CuTe, or CuTeDSL work.

Triton and Gluon development

Use the local Triton checkout as the primary source for APIs and implementation patterns. Prefer current tutorials and source over remembered signatures because Triton and Gluon evolve quickly.

Locate the checkout

Resolve the directory containing this SKILL.md, then use its repos/triton/ child. The installer links that path to agent-gpu-skills/third_party/triton/ or to the checkout supplied through TRITON_REPO.

In commands below, replace TRITON_REPO with the resolved absolute path:

bash

TRITON_REPO=/absolute/path/to/triton-skill/repos/triton

If the checkout is missing, run this from the agent-gpu-skills repository and reinstall the Skill:

bash

bash update-repos.sh triton
bash install.sh --skill triton-skill

Choose the source surface

Task	Start here
Triton language syntax and introductory patterns	`python/tutorials/`
Gluon layout and architecture-level patterns	`python/tutorials/gluon/`
Complete example kernels	`python/examples/`
Production matmul, reduction, top-k and SwiGLU	`python/triton_kernels/triton_kernels/`
`tl.*` definitions and semantics	`python/triton/language/`
JIT, autotuning and runtime behavior	`python/triton/runtime/`
Python compiler entry points	`python/triton/compiler/`
Triton and GPU dialect definitions	`include/triton/Dialect/`
Compiler analyses, transforms and lowering	`lib/`

Read quick-reference.md when choosing a tutorial, a complete example, or a production-kernel implementation.

Query workflow

Identify whether the task is Triton, Gluon, or compiler internals.
Find the closest current tutorial or implementation for the operation and architecture.
Verify every API used against its definition or another current call site.
Preserve the target workload's shape, dtype, stride, layout, masking, and numerical contract.
Establish correctness before changing launch parameters or optimization strategy.

Discover current examples before relying on a remembered filename:

bash

find "$TRITON_REPO/python/tutorials" -maxdepth 2 -type f | sort
find "$TRITON_REPO/python/examples" -type f | sort

Query Triton language usage and definitions:

bash

rg -n 'tl\.dot|tl\.dot_scaled' "$TRITON_REPO/python/tutorials"
rg -n '@triton\.autotune' "$TRITON_REPO/python/tutorials"
rg -n '^def (load|store|dot|dot_scaled)' \
  "$TRITON_REPO/python/triton/language"

Query Gluon architecture patterns:

bash

rg -n '@gluon\.jit' "$TRITON_REPO/python/tutorials/gluon"
rg -n 'wgmma|tcgen05|mbarrier|tma' \
  "$TRITON_REPO/python/tutorials/gluon" \
  "$TRITON_REPO/python/examples"

Trace production kernels:

bash

rg -n 'persistent|TensorDescriptor' \
  "$TRITON_REPO/python/triton_kernels/triton_kernels/matmul_details"

rg -n 'mxfp|flexpoint' \
  "$TRITON_REPO/python/triton_kernels/triton_kernels/numerics_details"

Trace compiler definitions and lowering:

bash

rg -n 'def.*Op' "$TRITON_REPO/include/triton/Dialect/Triton/IR"
rg -n 'Encoding' "$TRITON_REPO/include/triton/Dialect/TritonGPU/IR"
rg -n 'wgmma|tma|tcgen05' \
  "$TRITON_REPO/include/triton/Dialect/TritonNvidiaGPU"
rg -n 'Pattern|Rewrite' "$TRITON_REPO/lib/Conversion/TritonGPUToLLVM"

Show full SKILL.md (167 more words)Show less

Implementation discipline

Keep these layers separate during diagnosis:

text

Python kernel and launch metadata
  → Triton/Gluon IR and compiler transforms
  → generated GPU code on the selected target

A source-level pattern does not prove that the compiled kernel uses the intended instruction or memory path. Inspect compiler output or profile data when that distinction matters. Add cuda-skill for PTX semantics, compute capability, Nsight, or Compute Sanitizer details.

For correctness work:

compare against an independent reference over representative and boundary shapes;
test masked tails, non-power-of-two dimensions, strides and dtype conversions;
separate compilation failures from runtime correctness and numerical tolerance;
reproduce the original dispatch and launch metadata before reducing the case.

For performance work:

freeze the benchmark shape set and measurement method;
confirm autotune keys cover every dimension that changes the best configuration;
change one tile, warp, stage, persistence, or specialization choice at a time;
verify the generated path before attributing a result to TMA, WGMMA, tcgen05, or warp specialization.

Updating the source

From the agent-gpu-skills repository:

bash

bash update-repos.sh triton
python3 scripts/validate_repo.py --require-sources

The checkout follows Triton main, while third_party/UPSTREAMS.toml records the commit last accepted by this Skill. Review source-map drift before updating that record.

Files

SKILL.md and 1 other file in skills/triton-skill of slowlyC/agent-gpu-skills.

SKILL.md
quick-reference.md

Open the folder on GitHubat commit ae02d07

Compare with similar skills

Triton Skill next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Triton Skill compared with similar skills
Skill	Stars	Used in	Tokens	Auto-check	Licence	Repo updated
Triton Skill this skillslowlyC/agent-gpu-skills	169	—	~1.3k	Automated safety check: Pass	MIT	2 mo ago
Hyperpod Version Checkerawslabs/agent-plugins	916	—	~910	Automated safety check: Pass	Apache-2.0	today
Jetson Video SetupNVIDIA/skills	3.6k	1 repos	~2.4k	Automated safety check: Notes	Apache-2.0	today
Megatron-LM on SLURMNVIDIA/Megatron-LM	18k	—	~1.8k	Automated safety check: Pass	Apache-2.0	today
Optimize For GPUK-Dense-AI/scientific-agent-skills	48k	1 repos	~3.4k	Automated safety check: Pass	MIT	5 days ago
Paddle Design CompilerPaddlePaddle/Paddle	24k	—	~3.6k	Automated safety check: Pass	Apache-2.0	today

Similar skills

Hyperpod Version Checker
awslabs/agent-plugins
Official
Check and compare software component versions on SageMaker HyperPod cluster nodes - NVIDIA drivers, CUDA toolkit, cuDNN, NCCL, EFA, AWS OFI NCCL, GDRCopy, MPI, Neuron SDK (Trainium/Inferentia)…
916 GitHub stars~910 tokensUpdated today
AI & LLM EngineeringAuto-check passed
Jetson Video Setup
NVIDIA/skills
Official
A skill your agent uses when installing, repairing, reusing, inspecting, or verifying readiness of the native NVIDIA Video Codec SDK or PyNvVideoCodec on Jetson, including the one-frame…
3.6k GitHub starsUsed in 1 repo~2.4k tokens
AI & LLM EngineeringAuto-check: notes
Megatron-LM on SLURM
NVIDIA/Megatron-LM
Official
Shows how to launch distributed Megatron-LM training on a SLURM cluster: sbatch skeleton, torch.distributed.run setup, CUDA_DEVICE_MAX_CONNECTIONS rules and failure diagnosis.
18k GitHub stars~1.8k tokensUpdated today
AI & LLM EngineeringAuto-check passed
Optimize For GPU
K-Dense-AI/scientific-agent-skills
GPU-accelerates scientific Python on NVIDIA hardware and verifies that the result is correct and faster.
48k GitHub starsUsed in 1 repo~3.4k tokens
Data & AnalyticsAuto-check passed
Paddle Design Compiler
PaddlePaddle/Paddle
A skill your agent uses when working with Paddle 3.0 compiler full pipeline: SOT (Symbolic Opcode Translator) for bytecode-level dy2st graph capture, PIR (Paddle IR) for SSA-based intermediate…
24k GitHub stars~3.6k tokensUpdated today
AI & LLM EngineeringAuto-check passed
Graphsignal
graphsignal/graphsignal
Profile AI inference workloads (vLLM, SGLang, TensorRT-LLM, PyTorch, any GPU application) with the Graphsignal profiler and read the results from its local /signals JSON endpoint.
257 GitHub stars~6.3k tokensUpdated today
AI & LLM EngineeringAuto-check passed

Questions about Triton Skill

What does Triton Skill do?

Write, debug, and optimize Triton and Gluon GPU kernels from local upstream tutorials, production kernels, language definitions, and compiler source. Triton Skill is an agent skill from slowlyC/agent-gpu-skills. Write, debug, and optimize Triton and Gluon GPU kernels from local upstream tutorials, production kernels, language definitions, and compiler source.

When should I use Triton Skill?

Triton Skill fits situations like: the task explicitly involves triton.jit; triton.language; tensorDescriptor; triton autotune.

How do I install Triton Skill in Claude Code?

Run `npx skills add slowlyC/agent-gpu-skills --skill triton-skill -a claude-code`. Or copy the skill folder (skills/triton-skill in slowlyC/agent-gpu-skills) into .claude/skills/triton-skill in your project. Claude Code loads it when a task matches its description.

How do I install Triton Skill in Codex?

Run `npx skills add slowlyC/agent-gpu-skills --skill triton-skill -a codex`. Or copy the skill folder (skills/triton-skill in slowlyC/agent-gpu-skills) into .agents/skills/triton-skill in your project. Codex loads it when a task matches its description.

Can I use Triton Skill in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add slowlyC/agent-gpu-skills --skill triton-skill -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/triton-skill, .gemini/skills/triton-skill, .github/skills/triton-skill and .opencode/skills/triton-skill in your project.

What does Triton Skill need to run?

Going by SKILL.md and its folder, Triton Skill needs the command-line tools its instructions call (rg, bash and python3). Our summary lists: Python 3.

Does Triton Skill access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Triton Skill safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Triton Skill use?

Triton Skill is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Triton Skill use?

About 1.3k tokens (SKILL.md is roughly 5.3k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Triton Skill?

Skills that share tags, products or a category with Triton Skill: Hyperpod Version Checker (awslabs/agent-plugins, 916 stars), Jetson Video Setup (NVIDIA/skills, 3.6k stars), Megatron-LM on SLURM (NVIDIA/Megatron-LM, 18k stars) and Optimize For GPU (K-Dense-AI/scientific-agent-skills, 48k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Triton Skill?

slowlyC (a GitHub user) maintains it in slowlyC/agent-gpu-skills, which has 169 GitHub stars. The repository holds 4 skills in this directory. The repository was last updated on August 8, 2026.

Source: slowlyC/agent-gpu-skills on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.

Triton Skill

Install Triton Skill

Facts

At a glance

What it does

When your agent uses it

Example prompts

Requirements

Workflow steps

What it can do on your machine

Tool permissions

Runs code

Network

Credentials

Context cost

Safety

SKILL.md

Triton and Gluon development

Locate the checkout

Choose the source surface

Query workflow

Implementation discipline

Updating the source

Files

Compare with similar skills

Similar skills

Hyperpod Version Checker

Jetson Video Setup

Megatron-LM on SLURM

Optimize For GPU

Paddle Design Compiler

Graphsignal

More from slowlyC/agent-gpu-skills

Cutlass Skill

Tilelang Skill

Cuda Skill

Questions about Triton Skill

Install Triton Skill

Facts

At a glance

What it does

When your agent uses it

Example prompts

Requirements

Workflow steps

What it can do on your machine

Tool permissions

Runs code

Network

Credentials

Context cost

Safety

SKILL.md

Triton and Gluon development

Locate the checkout

Choose the source surface

Query workflow

Implementation discipline

Updating the source

Files

Compare with similar skills

Similar skills

More from slowlyC/agent-gpu-skills

Related

Questions about Triton Skill