Find the skill for your next task.
Browse reusable Agent Skills, each with a clear purpose and practical guidance.
add-shape-inference
Add or update type and shape inference for an ONNX operator. Use when asked to implement TypeAndShapeInferenceFunction, propagate shapes, add shape inference tests, fix shape inference bugs, or handle broadcasting logic.
onnxtxt
Read or write ONNX text format ("onnxtxt"). Use when authoring `.FunctionBody(R"ONNX(...)")` blocks, writing tests with `onnx.parser.parse_model` / `parse_graph`, using the C++ `OnnxParser`, debugging parser errors, or interpreting `Constant <value = ...>` and body-subgraph syntax.
resolve-backport
Resolve merge conflicts in Vitess backport PRs. Use when asked to resolve a backport, fix backport conflicts, or given a backport PR URL/number.
add-config-param
Step-by-step guide for adding a new configuration parameter to NetBox, covering both static parameters (settings.py) and dynamic parameters (database-backed, editable via the admin UI). Use when the user asks to add a new configuration option, setting, or parameter to NetBox.
contrib-op-shape-inference-memory-safety
Audit and fix out-of-range output writes in ONNX Runtime operator shape-inference functions. Use when reviewing or fixing a contrib (or standard) op TypeAndShapeInference where a getNumOutputs() guard precedes a write to a higher output index - optional trailing outputs make a smaller output count schema-valid, so getOutputType(index) can run one past the declared outputs at Graph::Resolve.
cuda-attention-kernel-patterns
Patterns and pitfalls for the ONNX domain Attention operator (opset 23/24) CUDA implementation. Use when modifying the dispatch cascade in core/providers/cuda/llm/attention.cc, writing mask/bias CUDA kernels, debugging attention test routing, or adding features to the ONNX Attention op. NOT for contrib domain MultiHeadAttention/GroupQueryAttention.
cuda-cutlass-fmha-incremental-rebuild
Use when rebuilding ONNX Runtime CUDA after editing CUTLASS fused-MHA headers (onnxruntime/contrib_ops/cuda/bert/cutlass_fmha/*.h such as kernel_forward.h or fmha_launch_template.h), or when a header edit "passed" an incremental build but test behavior did not change. Explains the nvcc depfile gotcha that produces stale Memory-Efficient-Attention (MEA) kernels and binaries, and how to force a correct recompile. Also covers disk-space frugality on shared GPU dev boxes.