triton-syntax

triton-syntax is a skill for Claude Code, Codex from mindspore-ai/akg. It costs 21 tokens per session (2,813 once invoked), scanned A, original, Apache-2.0.

A guide to Triton, a Python-like language for writing GPU programs in blocks of data. It explains its program model, data types, masking, and compiler-managed optimizations.

In plain words
What is it for?
Use it to implement operations such as element-wise addition, work with one- and two-dimensional tensors, handle boundaries, and choose numeric types.
Why use it?
It helps developers write GPU kernels in a more familiar style while handling common memory and scheduling details through Triton’s compiler.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one.

Good fit Use it to implement operations such as element-wise addition, work with one- and two-dimensional tensors, handle boundaries, and choose numeric types.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/mindspore-ai/akg/triton-syntax
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.

Any agent
npx skills add mindspore-ai/akg --skill triton-syntax
Clone the repo
git clone --depth 1 https://github.com/mindspore-ai/akg

Made for: Claude Code, Codex.

Wrote 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.

agentmods badge for triton-syntax

README.md
[![agentmods](https://agentmods.dev/badge/skills/mindspore-ai/akg/triton-syntax/github.svg)](https://agentmods.dev/skills/mindspore-ai/akg/triton-syntax)
Your own site
<a href="https://agentmods.dev/skills/mindspore-ai/akg/triton-syntax"><img src="https://agentmods.dev/badge/skills/mindspore-ai/akg/triton-syntax/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for triton-syntax

Your own site · 80×15
<a href="https://agentmods.dev/skills/mindspore-ai/akg/triton-syntax"><img src="https://agentmods.dev/badge/skills/mindspore-ai/akg/triton-syntax.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 21 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,813 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe. Third-party audits
  • NVIDIA SkillSpector pass 7 Sept 2026
How audits are shown
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 5.1 $0.00021 $0.02813
Opus 5.5 $0.00008 $0.01125
Sonnet 5.5 $0.00004 $0.00563
Haiku 4.5 $0.00002 $0.00281

Measured 2mo ago against content hash 88eb25d4694e, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

triton-syntax 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 2mo 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.

akg_agents/examples/run_skill/skills/triton-syntax/SKILL.md · 404 lines

How it starts

The opening of the file, as written. The whole thing — 404 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Triton编程语言

概述

Triton是一种专为GPU编程设计的Python DSL,目标是让GPU编程像NumPy一样简单。

核心特性

1. Python-like语法

@triton.jit
def add_kernel(
    x_ptr, y_ptr, output_ptr,
    n_elements,
    BLOCK_SIZE: tl.constexpr
):
    # 程序ID
    pid = tl.program_id(axis=0)
    
    # 块偏移
    block_start = pid * BLOCK_SIZE
    offsets = block_start + tl.arange(0, BLOCK_SIZE)
    
    # Mask处理边界
    mask = offsets < n_elements
    
    # 加载数据
    x = tl.load(x_ptr + offsets, mask=mask)
    y = tl.load(y_ptr + offsets, mask=mask)
    
    # 计算
    output = x + y
    
    # 存储结果
    tl.store(output_ptr + offsets, output, mask=mask)

2. 自动优化

Triton编译器自动进行:

  • 内存合并优化
  • 共享内存管理
  • 寄存器分配
  • 指令调度

3. Block编程模型

# 每个program处理一个block的数据
BLOCK_SIZE = 1024
grid = lambda meta: (triton.cdiv(n_elements, meta['BLOCK_SIZE']),)

add_kernel[grid](x, y, output, n_elements, BLOCK_SIZE=BLOCK_SIZE)

数据类型

基本类型

tl.float32    # 32位浮点
tl.float16    # 半精度浮点
tl.bfloat16   # Brain浮点
tl.int32      # 32位整数
tl.int64      # 64位整数

Tensor类型

# 1D tensor
x = tl.arange(0, BLOCK_SIZE)  # shape: [BLOCK_SIZE]

# 2D tensor
rows = tl.arange(0, BLOCK_M)[:, None]  # shape: [BLOCK_M, 1]
cols = tl.arange(0, BLOCK_N)[None, :]  # shape: [1, BLOCK_N]

内存操作

加载(Load)

# 基本加载
data = tl.load(ptr + offsets)

# 带mask加载(处理边界)
data = tl.load(ptr + offsets, mask=mask, other=0.0)

# 带cache hint
data = tl.load(ptr + offsets, cache_modifier=".ca")  # cache all

存储(Store)

# 基本存储
tl.store(ptr + offsets, data)

# 带mask存储
tl.store(ptr + offsets, data, mask=mask)

计算操作

算术运算

# 逐元素运算
c = a + b
c = a * b
c = a / b
c = tl.exp(a)
c = tl.log(a)
c = tl.sqrt(a)

规约操作

# Sum
total = tl.sum(x, axis=0)

# Max
maximum = tl.max(x, axis=0)

# Min
minimum = tl.min(x, axis=0)

矩阵运算

# Dot product
c = tl.dot(a, b)  # 高度优化的矩阵乘法

# 支持混合精度
c = tl.dot(a.to(tl.float16), b.to(tl.float16), acc=tl.float32)

Read the full file on GitHub · 404 lines

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. 2mo ago First seen · 404 lines · 21 tokens per session scan A 88eb25d4694e

Subscribe to this mod's changes

triton-syntax is a skill published in the GitHub repository mindspore-ai/akg (264 stars, last pushed yesterday), licensed Apache-2.0. It adds 21 tokens to every session and 2,813 once invoked, about $0.0001 per session on Opus 5.5. 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.