anthony-maio

Triton Persistent Warp Matmul — AI skill for Claude Code

AI community

Teach an AI agent to implement persistent, warp-specialized matmul kernels in Triton using TMA and producer/consumer warps.

How to install Triton Persistent Warp Matmul

Installs to ~/.claude/skills/anthony-maio-triton-skills-triton-persistent-warp-matmul/SKILL.md

Terminal

mkdir -p ~/.claude/skills/anthony-maio-triton-skills-triton-persistent-warp-matmul && curl -fsSL https://raw.githubusercontent.com/anthony-maio/triton-skills/HEAD/skills/triton-persistent-warp-matmul.md -o ~/.claude/skills/anthony-maio-triton-skills-triton-persistent-warp-matmul/SKILL.md

Restart Claude Code, or start a new session, for it to be picked up.

What Triton Persistent Warp Matmul does

name: triton-persistent-warp-matmul description: Teach an AI agent to implement persistent, warp-specialized matmul kernels in Triton using TMA and producer/consumer warps.

Persistent & Warp-Specialized Matmul Kernels in Triton

**Targets:** Triton >= 3.0; TMA/warp specialization requires SM90+ (Hopper)

Overview This skill teaches how to implement a persistent GEMM in Triton where fewer thread blocks than output tiles are launched and each block iterates over multiple tiles. It cov

Alternatives in AI

Osaurus — Own your AI. The native macOS harness for AI agents -- any model, persistent memory, autonomous execution, cry 5.1k ★
Agentmemory — #1 Persistent memory for AI coding agents based on real-world benchmarks 1.9k ★
Azure AI Agents .net — Persistent AI agents with threads and tools 1.8k ★