Routes Quark ONNX user requests to the smallest correct downstream workflow.
npx skills add https://github.com/amd/Quark --skill quark-onnx-router
Please help me install this Agent Skill. Skill: quark-onnx-router Source: https://github.com/amd/Quark/tree/main/.claude/skills-impl/l1-atomic/onnx/quark-onnx-router Command: npx skills add https://github.com/amd/Quark --skill quark-onnx-router
This Skill routes Quark ONNX user requests to the correct atomic workflow so users do not waste time on the wrong installation, analysis, planning, execution, or validation step.
Use the quark-onnx-router skill to interpret my ONNX quantization goal and tell me which Quark skill should run first.
High-intent search queries and answers about installing and using this skill.
ONNX model quantization starts by classifying your intent from a plain-language request and routing it to the correct downstream workflow. The router gathers required hardware facts and workspace paths before selecting the smallest correct skill.
Routing a YOLO ONNX model for Ryzen AI involves classifying the quantization goal and directing it into the appropriate PTQ workflow. The router enforces guardrails to prevent cross-backend mixing and ensures prerequisite facts are collected.
ONNX Runtime setup requires backend disambiguation to ensure correct model intake. The router determines your specific runtime environment and gathers necessary workspace paths before handing off to the smallest correct installation or inspection workflow.
Checking model compatibility before a Quark ONNX PTQ workflow is supported through preflight routing. The router directs your model inspection request to the correct atomic skill, ensuring all prerequisite hardware facts are validated first.
ONNX quantization workflows fail without prerequisite fact gathering because guardrails prevent execution when required context is missing. The router enforces collection of hardware facts and workspace paths before allowing handoff into any downstream execution skill.
Validating a quantized ONNX output after execution is handled by routing your validation request to the correct atomic skill. The router interprets your result checking intent and directs it to the specific validation workflow in the Quark ONNX flow.
amdExtracts ONNX model metadata needed for Quark PTQ planning and compatibility assessment.
amdQuantize PyTorch transformer models into deployment artifacts using AMD Quark PTQ workflows.
amdVerify Quark ONNX skill routing, planning, artifacts, and recovery checks.
galaxyprojectRoute Nextflow conversion tasks to the appropriate Galaxy sub-skill.
