About

AMD's verified Agent Skills in one plugin: route image/audio through local AI on Ryzen AI, serve LLMs on AMD Instinct GPUs with vLLM, optimize inference throughput with Hyperloom, and analyze GPU kernel and PyTorch trace performance.

Components

• Skills — ./local-ai-use, ./local-ai-app-integration, ./serving-llms-on-instinct, ./tracelens-analysis-orchestrator, ./hyperloom-workload-optimizer, local-ai-use, local-ai-app-integration, serving-llms-on-instinct, tracelens-analysis-orchestrator, hyperloom-workload-optimizer

Declaration (derived by the registry from the source — the publisher wrote no manifest)

• component Provides 5 skill(s): local-ai-use, local-ai-app-integration, serving-llms-on-instinct, tracelens-analysis-orchestrator, hyperloom-workload-optimizer
• component Skill "local-ai-use": >- Makes this agent generate images, transcribe audio, and synthesize speech on the user's own machine through a local Lemonade Server instead of a paid cloud API. Use it above all to change that routing persistently, from now on — keep gen
• component Skill "local-ai-app-integration": >- Integrates local AI capabilities into applications using Embeddable Lemonade. Use when the user wants to add local AI, offline AI, private AI, on-device AI, a local LLM, local chat, embeddings, image generation, speech-to-text, or text-t
• component Skill "serving-llms-on-instinct": >- Serves AI models on AMD Instinct GPU hardware using vLLM. Use this skill whenever the user wants to run, serve, deploy, start, host, or launch a language model on an AMD GPU, AMD Instinct, MI300X, MI325X, MI350X, or MI355X. Also use when
• component Skill "tracelens-analysis-orchestrator": >- Orchestrates modular PyTorch profiler trace analysis with TraceLens: generates perf reports, prepares category data, runs system-level and compute-kernel subagents in parallel, validates outputs, and writes a prioritized stakeholder repo
• component Skill "hyperloom-workload-optimizer": >- Autonomously optimizes end-to-end LLM inference throughput on AMD Instinct GPUs and reports a validated gain, using the Hyperloom multi-agent optimizer. Given a model, framework, workload (TP/EP, concurrency, ISL/OSL, precision), an obje
• undeclared Undeclared: No plugin.json was reachable at the source — declaration derived from the marketplace entry alone; tool access is undeclared

Undeclared — sandbox / Playground only until the publisher confirms

• No plugin.json was reachable at the source — declaration derived from the marketplace entry alone; tool access is undeclared

Derived from: marketplace entry in anthropics/claude-plugins-official (no plugin.json reachable) · SKILL.md frontmatter (local-ai-use) · SKILL.md frontmatter (local-ai-app-integration) · SKILL.md frontmatter (serving-llms-on-instinct) · SKILL.md frontmatter (tracelens-analysis-orchestrator) · SKILL.md frontmatter (hyperloom-workload-optimizer)

Validation

Files hashed

Use this skill — no import, no export

In Claude Code, once:

Then install by name:

Or ask any connected agent:

Attestation

The registry signs the hash of the source at the pinned ref. If the source changes after pinning, the hash fails and integrating hosts refuse it.

Record