Intel Agent Skills

intel/skills

Skills for running, porting, profiling, and tuning code on Intel CPUs, GPUs, and the libraries and runtimes behind them. Find one below and install it with a single command.

npx skills add intel/skills

Search and filter skills

Hardware
Products

cuda-to-xpu-migration

  • Hardware class: gpu
  • Intel product: Intel Arc
  • Intel product: Intel Arc Pro

Create a CUDA-to-XPU migration assessment for an existing AI repo. Identify CUDA-specific assumptions, route to the right XPU skills, produce a migration report. Use when the user has a CUDA repo, notebook, Dockerfile, launch script, HF / vLLM / SGLang workload, or Triton kernel and asks to migrate it to Intel Arc / Arc Pro / Battlemage / XPU — including "convert this to XPU" / "move it to XPU" and the bare "migrate this repo" request where scope is not yet set. A request that says "port" routes to xpu-port. Plans and routes only. Not for executing an already-scoped rewrite (use xpu-port), running migrated code, or measuring it.

Install this skill

npx skills add intel/skills --skill cuda-to-xpu-migration

Source

View source on GitHub

dpnp-interop

  • Hardware class: gpu
  • Intel product: dpnp

Passing dpnp arrays to and from other Python libraries on Intel CPUs and GPUs. Use when dpnp numeric work has to feed pandas, scikit-learn, PyTorch, or TensorFlow, when one of those libraries raises a type error on a dpnp array, when a pipeline mixes device math with host-only libraries, or when the user asks where in a pipeline the conversion belongs. Covers the boundary conversion pattern per library, the Intel extensions that accelerate the host side, and why a conversion inside a loop erases the benefit.

Install this skill

npx skills add intel/skills --skill dpnp-interop

Source

View source on GitHub

dpnp-io

  • Hardware class: gpu
  • Intel product: dpnp

Reading and writing files from dpnp code on Intel CPUs and GPUs. Use when the user needs to load an array into dpnp or save a dpnp result — .npy, .npz, HDF5 via h5py, Zarr, CSV or plain text — when a file is larger than device memory and has to be read in chunks, or when they ask why dpnp has no save function of its own. Covers the NumPy conversion round trip, chunked and incremental patterns, and choosing a format by dataset size.

Install this skill

npx skills add intel/skills --skill dpnp-io

Source

View source on GitHub

dpnp-linalg-fft

  • Hardware class: gpu
  • Intel product: dpnp
  • Intel product: oneMKL

Linear algebra and FFT with dpnp on Intel CPUs and GPUs, backed by oneMKL. Use when a matrix multiply, solve, decomposition, eigenvalue problem, or Fourier transform is the hot part of NumPy code, when the user asks whether dpnp covers a linalg or FFT call, when an FFT result differs slightly from NumPy's, or when a large transform runs out of device memory. Covers the supported surface, transform sizing and plan reuse, fallbacks for what is missing, and how to check conditioning before solving.

Install this skill

npx skills add intel/skills --skill dpnp-linalg-fft

Source

View source on GitHub

dpnp-memory

  • Hardware class: gpu
  • Intel product: dpnp
  • Intel product: dpctl

Device memory management for dpnp arrays on Intel CPUs and GPUs. Use when a dpnp script grows in memory until it fails, when a dataset does not fit in device memory, when an array turns out to be on a different device than expected, or when a loop allocates a new array on every iteration. Covers USM allocation, inspecting placement and queues with dpctl, reusing an output buffer, chunking a workload larger than the device, and the tools that report device memory use.

Install this skill

npx skills add intel/skills --skill dpnp-memory

Source

View source on GitHub

dpnp-migration

  • Hardware class: gpu
  • Intel product: dpnp

Porting an existing NumPy or CuPy program to dpnp on Intel CPUs and GPUs. Use when deciding whether a codebase can run on dpnp at all, when a call raises NotImplementedError or AttributeError after the import was swapped, when the user asks whether dpnp supports a specific NumPy function or family, or when CuPy code has to move to Intel hardware. Covers probing the installed release for what it actually implements, the families that have no device counterpart, the fallback wrapper for the ones that do not, and where CuPy's device model differs from dpnp's.

Install this skill

npx skills add intel/skills --skill dpnp-migration

Source

View source on GitHub

dpnp-quickstart

  • Hardware class: gpu
  • Intel product: dpnp
  • Intel product: dpctl

NumPy-compatible array operations optimized for Intel hardware. Use when the user wants to migrate or port NumPy code to dpnp, asks whether a NumPy hot path can run on an Intel CPU or GPU, needs to check dpnp installation or SYCL device selection with dpctl, hits a NumPy API dpnp does not implement, or wants to compare dpnp against NumPy. Covers install, device control, fallback patterns, and profiling.

Install this skill

npx skills add intel/skills --skill dpnp-quickstart

Source

View source on GitHub

dpnp-random

  • Hardware class: gpu
  • Intel product: dpnp
  • Intel product: oneMKL

Random number generation with dpnp on Intel CPUs and GPUs, backed by oneMKL. Use when NumPy random calls move to dpnp, when a seeded dpnp run does not reproduce a NumPy sequence, when a distribution turns out not to be implemented, when results have to be reproducible across machines, or when random data feeds a training or augmentation loop. Covers the supported distributions, what seeding does and does not guarantee, the host fallback, and where to generate data so it does not bounce between host and device.

Install this skill

npx skills add intel/skills --skill dpnp-random

Source

View source on GitHub

dpnp-troubleshooting

  • Hardware class: gpu
  • Intel product: dpnp
  • Intel product: dpctl

Diagnosing dpnp failures on Intel CPUs and GPUs. Use when dpnp raises NotImplementedError or an unexpected TypeError, when the import fails or a SYCL runtime library is missing, when no SYCL device is visible, when dpctl reports a device the user did not expect, or when dpnp code runs slower than the NumPy it replaced. Covers the fallback pattern for unimplemented APIs, install repair, forcing CPU execution, and the handoff to libraries that only accept NumPy arrays.

Install this skill

npx skills add intel/skills --skill dpnp-troubleshooting

Source

View source on GitHub

linux-perf

  • Hardware class: cpu
  • Intel product: linux-perf

Profile and fix Linux performance problems using perf. Workflows: (A) hardware counters -- IPC, cache-miss, branch mispredictions; (B) hotspot profiling -- which functions and source lines consume CPU, with SIMD and accumulator detection; (C) cache-line contention -- false sharing, HITM, perf c2c; (D) core-count scaling -- dual-profile comparison, bottleneck categorization; (E) structured hotspot report with annotated source and pattern observations. Resolution strategies: TTAS spinlock, SIMD upconversion, parallel accumulator, structured false-sharing fix, per-CPU stats. Trigger on: perf, profiling, profile, hotspot, hotspots, cache miss, IPC, false sharing, HITM, scaling, core count, thread scaling, bottleneck, slow code, CPU bound, why is this slow, where does time go, does not scale. When in doubt, invoke this skill -- better to use it unnecessarily than to miss a performance opportunity.

Install this skill

npx skills add intel/skills --skill linux-perf

Source

View source on GitHub

llamacpp-xpu-run

  • Hardware class: gpu
  • Intel product: Intel GPU
  • Intel product: Level Zero

Run a GGUF model on an Intel GPU using llama.cpp's SYCL backend (Level Zero) with the official intel.Dockerfile. Covers building the Docker image from source at a pinned tag, launching llama-server with an OpenAI-compatible API, device selection, multi-GPU layer splitting, all recommended runtime env vars, flash-attention, and quantisation selection. Use when the user has a GGUF model and wants fast local inference or an OpenAI-compatible endpoint on Intel GPU without Python/PyTorch. The CUDA analogue is llama.cpp built with -DGGML_CUDA=ON. Use vllm-xpu-run instead for safetensors models with continuous batching at scale; use torch-xpu-run for Hugging Face Transformers direct.

Install this skill

npx skills add intel/skills --skill llamacpp-xpu-run

Source

View source on GitHub

mkl-extension-advisor

  • Hardware class: cpu
  • Intel product: mkl_fft
  • Intel product: mkl_random
  • Intel product: mkl_umath
  • Intel product: mkl-service

Deciding whether Intel's MKL extension packages apply to NumPy or SciPy code on Intel CPUs. Use when a user asks whether mkl_fft, mkl_random, or mkl_umath help their code, or points at a snippet, function, file, or codebase using np.fft, scipy.fft, np.random, or element-wise math ufuncs. Also use to check whether these extensions are already active in an environment, to fix an install so they and the SciPy FFT backend actually work, or to judge whether patching would change results and break exact-output tests. DO NOT use for GPU work, non-Intel CPUs, or mkl-service thread tuning.

Install this skill

npx skills add intel/skills --skill mkl-extension-advisor

Source

View source on GitHub

model-can-it-fit

  • Hardware class: gpu
  • Intel product: Intel GPU

Estimate whether a Hugging Face decoder-only LLM, MoE, or VLM fits in Intel GPU VRAM for a quantization, context length, concurrency, runtime, and tensor-parallel setting. Use for memory-fit or max-model-len planning before launch. Reports weights, KV cache, activations, framework overhead, and first mitigation. Not for diffusion. Memory-only — does NOT predict throughput, tokens/sec, latency, or runtime config; route those to bench/deploy/recommend skills.

Install this skill

npx skills add intel/skills --skill model-can-it-fit

Source

View source on GitHub

model-config-recommend

  • Hardware class: gpu
  • Intel product: Intel Arc

Recommend a vLLM-XPU deployment config (quant, KV dtype, DP/TP, max concurrency, max context) for a Hugging Face decoder-only LLM on Intel Arc B-series GPUs using roofline math against published hardware specs. Experimental; predictions are physics-bounded ranges, not measured throughput. Use after xpu-discover and before vllm-xpu-run.

Install this skill

npx skills add intel/skills --skill model-config-recommend

Source

View source on GitHub

onetbb-quickstart

  • Hardware class: cpu
  • Intel product: oneTBB

Getting started with Intel oneTBB for C++ parallelism on Intel CPUs. Use when a C++ loop or reduction should run on multiple threads with oneTBB, when the user needs the headers, namespace, or CMake wiring for a first oneTBB program, when a parallel_for body has a data race, or when a reduction is accumulating into a shared variable. Covers parallel_for and parallel_reduce over blocked_range, the build setup, and the pitfalls of the task-based model.

Install this skill

npx skills add intel/skills --skill onetbb-quickstart

Source

View source on GitHub

performance-patterns

  • Hardware class: cpu

Detect and fix x86/C/C++ performance patterns from source code or profiling output (perf, VTune, flamegraphs). Invoke when the user asks to optimize, review for performance, or write new SIMD/vectorized code — even without profiling data. Trigger on: serial accumulator loops, narrow SIMD (xmm/ymm that could be ymm/zmm), _mm* intrinsics, HITM/cmpxchg clusters, false sharing, missing restrict or vzeroupper, futex_wake/notify_all thundering herd, hot symbol inside a system library (.so) with a version gap, or any request to write a fast reduction, dot product, or CPU-dispatched function. Patterns: serial accumulator, TTAS spinlock, SIMD upconversion (zipper), false sharing, per-CPU stats, missing vzeroupper, missing restrict, cv-thundering-herd, mutex-to-rwlock, CPU dispatch, library version upgrade, fast CRC32C, known algorithms (Cosine Similarity, Hamming Distance, Jaccard Distance), SIMD sort (x86-simd-sort).

Install this skill

npx skills add intel/skills --skill performance-patterns

Source

View source on GitHub

phoronix-test-suite

  • Hardware class: cpu

Install, run, parse, and optimize benchmarks from the Phoronix Test Suite (PTS). Use this skill whenever the user mentions "phoronix", "pts/", or "phoronix-test-suite", or asks to run, measure, improve, or optimize a PTS test — e.g., "run pts/mt-dgemm", "optimize pts/compress-zstd", "what score does pts/x265 get". Trigger immediately on any pts/<testname> reference, even if the user doesn't explicitly say "phoronix". Also trigger when the user asks to find or edit the source code of a PTS test.

Install this skill

npx skills add intel/skills --skill phoronix-test-suite

Source

View source on GitHub

sglang-xpu-bench

  • Hardware class: gpu
  • Intel product: Intel GPU

Benchmark a running SGLang-XPU server on an Intel GPU using sglang.bench_serving. Measures TTFT, TPOT, ITL, end-to-end latency, and throughput against the OpenAI-compatible endpoint. Use after sglang-xpu-run. Not for vLLM servers (use vllm-xpu-bench) or no-server PyTorch (use torch-xpu-bench).

Install this skill

npx skills add intel/skills --skill sglang-xpu-bench

Source

View source on GitHub

sglang-xpu-run

  • Hardware class: gpu
  • Intel product: Intel GPU
  • Intel product: Level Zero

Serve a Hugging Face safetensors model on an Intel GPU using SGLang's XPU backend with the OpenAI-compatible API. Covers pulling the pre-built intel/sglang-dev:latest image, fixing the render-group and UMD/kernel compatibility issues that affect non-root sglang images, the SYCL_UR / Level Zero env vars needed on Battlemage, the --device xpu --attention-backend intel_xpu flag set, multimodal serving, and how to validate output content (not just HTTP 200). Use when the user needs SGLang's RadixAttention prefix caching or grammar-constrained output; for broad-coverage serving on Intel today prefer vllm-xpu-run, and for benchmarking a running server use sglang-xpu-bench.

Install this skill

npx skills add intel/skills --skill sglang-xpu-run

Source

View source on GitHub

torch-xpu-bench

  • Hardware class: gpu
  • Intel product: Intel GPU

Benchmark a Hugging Face model on an Intel GPU through pure PyTorch + Transformers, single-process, no HTTP server. Measures generate() throughput in tokens/sec, time-to-first-token, decode-step latency, and peak XPU memory. Also covers diffusion and encoder-only models via references/non-llm-snippets.md. Use after model-can-it-fit to validate predicted memory against torch.xpu.max_memory_allocated().

Install this skill

npx skills add intel/skills --skill torch-xpu-bench

Source

View source on GitHub

torch-xpu-profile

  • Hardware class: gpu
  • Intel product: Intel GPU

Profile a Hugging Face model on Intel GPU at the PyTorch level with torch.profiler and Kineto. Captures CPU + XPU timeline, exports Chrome trace, identifies hottest kernels and async-overlap gaps. Use when the user asks why a model is slow, which op is the bottleneck, or where the GPU is idle. Not for profiling inside a running vLLM server (use vllm-xpu-profile) or for SYCL-kernel-level signal beneath the PyTorch op layer (use xpu-profile-unitrace).

Install this skill

npx skills add intel/skills --skill torch-xpu-profile

Source

View source on GitHub

torch-xpu-run

  • Hardware class: gpu
  • Intel product: Intel GPU

Run an arbitrary Hugging Face safetensors model on an Intel GPU using upstream PyTorch (>= 2.8) with the built-in torch.xpu device. Covers loading from the Hub, picking the right dtype, autocast, multi-GPU with accelerate's device_map, and the CUDA -> XPU code translation a user has to do once. Use for the Transformers / Accelerate / Diffusers path. Not for OpenAI-compatible serving (use vllm-xpu-run); explicitly not via intel-extension-for-pytorch (ipex) or ipex-llm — those paths are end-of-life and upstream PyTorch supersedes them.

Install this skill

npx skills add intel/skills --skill torch-xpu-run

Source

View source on GitHub

vllm-xpu-bench

  • Hardware class: gpu
  • Intel product: Intel GPU

Benchmark a running vLLM-XPU OpenAI-compatible server on an Intel GPU using vllm bench. Measures TTFT (time-to-first-token), TPOT (time-per-output-token), ITL (inter-token latency), end-to-end latency, and throughput under concurrency. Covers online (vllm bench serve) and offline (vllm bench throughput) modes; concurrency sweeps and quant comparison live in references/sweep-and-compare.md. Use after vllm-xpu-run when the user asks "how fast is this?".

Install this skill

npx skills add intel/skills --skill vllm-xpu-bench

Source

View source on GitHub

vllm-xpu-profile

  • Hardware class: gpu

Profile a running vLLM-XPU server with torch.profiler around a window of real requests, either via /start_profile and /stop_profile HTTP endpoints or via vllm bench --profile for offline runs. Use to find the dominant op under real concurrent traffic. Not for pure PyTorch (use torch-xpu-profile), SYCL kernel-level signal (use xpu-profile-unitrace), throughput numbers (use vllm-xpu-bench), or non-vLLM servers.

Install this skill

npx skills add intel/skills --skill vllm-xpu-profile

Source

View source on GitHub

vllm-xpu-run

  • Hardware class: gpu
  • Intel product: Intel GPU

Serve a Hugging Face safetensors model on an Intel GPU with upstream vLLM-XPU's OpenAI-compatible API. Covers image choice, container launch, the right vllm serve flags (dtype, enforce-eager, model-impl fallback, attention backend, quant + KV-cache pairing), and the transformers-backend fallback for unsupported architectures. Use for /v1/chat/completions or /v1/completions on an Intel GPU. Not for pure PyTorch without a server (use torch-xpu-run), throughput numbers (use vllm-xpu-bench), or NVIDIA (use vllm-project/vllm-skills).

Install this skill

npx skills add intel/skills --skill vllm-xpu-run

Source

View source on GitHub

xpu-container-run

  • Hardware class: gpu
  • Intel product: Intel GPU

Launch a Docker container with Intel GPU access on Linux. Encodes the correct combination of --device /dev/dri, render-group access, --ipc=host, ZE_AFFINITY_MASK pinning, Hugging Face cache mount, and --entrypoint /bin/bash for interactive use. Use when running any Intel-XPU container (vLLM-XPU, sglang-xpu, torch-XPU, llama.cpp SYCL, etc.) and the device must be visible inside. The CUDA analogue is docker run --gpus all — Intel has no --gpus flag, you pass the Direct Rendering Manager (DRM) nodes directly.

Install this skill

npx skills add intel/skills --skill xpu-container-run

Source

View source on GitHub

xpu-deploy-plan

  • Hardware class: gpu
  • Intel product: Intel XPU

Plan an end-to-end Intel XPU model deployment by chaining existing skills. Calls xpu-runtime-preflight (readiness), model-can-it-fit (sizing), model-config-recommend (flags), and the selected runtime skill (vllm-xpu-run / sglang-xpu-run / torch-xpu-run), then writes a single PLAN.md with one exact launch command, smoke test, and rollback to .out/skills/xpu-deploy-plan/. Use when the user asks for a coordinated plan (not a direct deploy/serve request) — wants the orchestration across preflight, fit, config, launch, smoke test, and rollback, or asks which skills to run and in what order.

Install this skill

npx skills add intel/skills --skill xpu-deploy-plan

Source

View source on GitHub

xpu-discover

  • Hardware class: gpu
  • Intel product: Intel Arc
  • Intel product: Intel Arc Pro
  • Intel product: Intel Data Center GPU Max

Inventory Intel GPUs (Arc, Arc Pro, Data Center GPU Max) on a Linux host. Detect devices, check driver health, list processes using each XPU, run a quick diagnostic, and read live utilisation.

Install this skill

npx skills add intel/skills --skill xpu-discover

Source

View source on GitHub

xpu-model-type-detect

  • Hardware class: gpu
  • Intel product: Intel XPU

Before loading a Hugging Face model on Intel XPU, detect its actual type (text generation, text encoder, seq2seq, vision classification, vision-language, audio encoder, audio seq2seq, multimodal VL, diffusion, time-series, reward model, masked LM) so the agent picks the right AutoModel class and input kwargs. Prevents "got unexpected keyword argument 'pixel_values'" and "empty logits" errors from mis-routing. Use before torch-xpu-run or vllm-xpu-run when the user gives a model id the agent hasn't seen before, or when a smoke test fails with a wrong-input signature.

Install this skill

npx skills add intel/skills --skill xpu-model-type-detect

Source

View source on GitHub

xpu-port

  • Hardware class: gpu

Execute a single-target CUDA-to-XPU port of a PyTorch repo with libcst-based scan, mechanical rewrite, and CPU FP64 vs target-dtype correctness verify on one forward pass. Use when the request says "port" — "port my repo to XPU", "port my repo at <path> to XPU", "rewrite the CUDA calls to XPU", "apply the mechanical transforms", "run the scan and rewrite", "make the port changes now". Not for the "migrate" verb ("migrate my repo", "migrate this repo to XPU") or a bare whole-repo workflow request where scope is not yet set — those start with cuda-to-xpu-migration, whose plan routes here. Not for assessment-only, throughput (torch-xpu-bench), op-level slowness (torch-xpu-profile), custom CUDA C++ extensions, or dual-target CUDA+XPU codebases.

Install this skill

npx skills add intel/skills --skill xpu-port

Source

View source on GitHub

xpu-profile-unitrace

  • Hardware class: gpu
  • Intel product: Intel XPU
  • Intel product: pti-gpu
  • Intel product: oneCCL

Profile Intel-XPU workloads at the SYCL / Level Zero kernel level via Intel pti-gpu's unitrace. Captures per-API-call and per-kernel timing, memory transfers, oneCCL / MPI events, and hardware counters PyTorch-level profilers cannot see. Use when a hot op is already known at the torch.profiler layer and the user needs the SYCL kernel beneath, or when profiling oneCCL collectives in multi-GPU runs. Not for PyTorch-level signal (use torch-xpu-profile / vllm-xpu-profile). Requires building unitrace from source.

Install this skill

npx skills add intel/skills --skill xpu-profile-unitrace

Source

View source on GitHub

xpu-runtime-preflight

  • Hardware class: gpu
  • Intel product: Intel GPU

Run a read-only go/no-go preflight before any Intel GPU/XPU skillpack work. Checks driver health, /dev/dri permissions, render/video groups, Docker, /dev/shm, disk, proxy, and optional container-level XPU visibility. Use when the user asks whether a machine is ready for XPU model work or needs a reusable lab readiness report. Not for launching workloads, pulling images, editing system config, or verifying model output.

Install this skill

npx skills add intel/skills --skill xpu-runtime-preflight

Source

View source on GitHub

xpu-system-setup

  • Hardware class: gpu
  • Intel product: Intel GPU
  • Intel product: Intel Arc Pro
  • Intel product: xpu-smi
  • Intel product: Level Zero

First-time setup for Intel XPU/GPU hosts. Detects what's missing and installs xpu-smi, configures user groups (render), sets up Intel GPU PPA repository, installs Level Zero runtime, installs Docker, and runs a post-setup verification gate. Prompts before each installation by default (use --auto for unattended). Also handles Battlemage (Arc Pro B60/B70) prerequisites on Ubuntu 24.04: nomodeset removal, OEM kernel upgrade, and compute runtime 26.18+ — use check_battlemage_prerequisites.sh when xpu-smi shows No device discovered or clinfo shows 0 platforms. Use when a bare-metal or minimally-configured machine needs to be prepared for XPU model work.

Install this skill

npx skills add intel/skills --skill xpu-system-setup

Source

View source on GitHub