Language:JapaneseEnglish

amd-skills Claude Code Plugin

Author
AMD
Category
Development
Topics
AI Agents & AI App Development · Claude Code Customization & Workflow · Monitoring & Observability
First cataloged
2026-07-17 (UTC)
Explanation last updated
2026-08-15 (UTC)
Source (GitHub) last updated
2026-08-28 (UTC) (2 days ago)

The explanation below is AI-generated. Please verify it against the sources.

The skills are grouped into three areas: client-side skills for running local AI on Ryzen AI hardware, cross-stack skills such as ROCm diagnostics and model request routing, and server-side skills for deploying and profiling large language models on AMD Instinct GPUs and EPYC CPUs. Each skill follows the standard Agent Skills format, built around a SKILL. Skills can be installed through a CLI command or by manually copying skill folders into an agent's skills directory. The README describes the project as an evolving technology preview released under the MIT license.

What you can do with amd-skills

  • Diagnose and fix ROCm, HIP, PyTorch, and llama. cpp configuration problems on AMD GPUs
  • Build a Lemonade model router that directs requests based on content or required capability
  • Deploy and benchmark LLM inference endpoints on AMD Instinct GPUs using vLLM
  • Evaluate GPU kernel correctness and performance and benchmark vLLM/SGLang inference
  • Orchestrate PyTorch profiler trace analysis and produce a prioritized stakeholder report

Sources

History of amd-skills

  • Claude Code Plugin Added amd-skills

Back to list