walkthroughs/README.md

# AMD Skills walkthroughs

## Requirements

This collection of walkthroughs assumes you already have [Claude Code](https://www.anthropic.com/claude-code) set up on your machine.

Participants using other agents are still encouraged to participate. You might have to adapt some of the instructions provided below to your specific agent.

Choose a skill to get started.

* [lemonade-router-builder](./lemonade-router-builder.md): Generate a valid Lemonade router policy JSON from a plain-English description of routing intent.
* [local-ai-use](./local-ai-use.md): Teach your agent how to run image generation locally.
* [local-ai-app-integration](./local-ai-app-integration.md): Add a local AI mode to a cloud-only app.
* [hyperloom-workload-optimizer](./hyperloom-workload-optimizer.md): Set up Hyperloom and autonomously optimize large language model (LLM) inference throughput on AMD Instinct GPUs.
* [magpie-kernel-evaluator](./magpie-kernel-evaluator.md): Benchmark inference, identify bottlenecks, and analyze, compare, and revalidate optimized GPU kernels with Magpie.
* [serving-llms-on-instinct](./serving-llms-on-instinct.md): Bring up a vLLM OpenAI-compatible LLM endpoint on an AMD Instinct GPU.
* [serving-llms-on-epyc](./serving-llms-on-epyc.md): Bring up a vLLM + zentorch LLM endpoint on an AMD EPYC CPU.
* [tracelens-analysis-orchestrator](./tracelens-analysis-orchestrator.md): Run agentic PyTorch profiler trace analysis and produce a prioritized performance report.