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.
1 skill
Deploy a Prometheus + Grafana + XPU Manager monitoring stack on an Intel GPU host using the intel/xe-full-stack-monitoring repo (plain Docker Compose). Brings up containers for real-time GPU, CPU, and vLLM metrics collection and visualization. Use when the user wants to monitor XPU utilization, temperature, power, memory, or vLLM inference performance.
1 skill
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.
1 skill
Estimate whether a Hugging Face decoder-only LLM, MoE, VLM, sliding-window, or hybrid state-space (SSM) model 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, recurrent state, 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.
1 skill
Recommend how to configure vLLM-XPU for a Hugging Face decoder-only LLM on Intel Arc B-series GPUs: choose quantization, KV dtype, DP/TP layout, max concurrency, and max context using roofline math against published hardware specs. Use when the user asks "How should I configure vLLM?", requests the best vLLM configuration, or asks how to use one or multiple Arc cards. Must run recommend.py; predictions are physics-bounded ranges, not measured throughput. Use after xpu-discover and before vllm-xpu-run.
1 skill
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).
1 skill
Serve a Hugging Face safetensors model on an Intel GPU using SGLang's XPU backend with the OpenAI-compatible API. Pins the validated `lmsysorg/sglang:v<version>-xpu` release image and flags newer unvalidated releases, then covers the container flag set for Intel DRM devices, the UMD/kernel pairing the image pins, the `--device xpu --attention-backend intel_xpu` flag set, page-size and quantization constraints, 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.
1 skill
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()`.
1 skill
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).
1 skill
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.
1 skill
Bring up a vLLM-Omni server (`vllm serve --omni`) on Intel Arc / Arc Pro B-series GPUs (XPU) for omni-modality and generative models — Qwen2.5-Omni / Qwen3-Omni chat, text-to-image (Qwen-Image, Z-Image, FLUX), video (Wan) and TTS (Qwen3-TTS). Covers building the image from upstream `docker/Dockerfile.xpu` on the matching `vllm/vllm-openai-xpu` base (no prebuilt XPU image or wheel exists), container flags, per-stage device placement with `--stage-overrides`, single-card layouts, and output-validated smoke tests for chat, image and speech endpoints. Use for any vLLM-Omni request on Intel XPU. Not for text-only LLM serving (use vllm-xpu-run).
1 skill
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?".
1 skill
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.
1 skill
Serve a Hugging Face safetensors model on an Intel GPU with upstream vLLM-XPU's OpenAI-compatible API, or check whether a model or architecture is currently documented on XPU. Covers live support lookup, image choice, container launch, known serve-flag requirements, model-impl fallback, and attention/quant compatibility. Use to launch /v1/chat/completions or /v1/completions, troubleshoot a launch, or check model support. Not for choosing the best quantization, KV dtype, DP/TP layout, context, or concurrency (use model-config-recommend).
1 skill
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.
1 skill
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 to deploy a model on an Intel GPU end-to-end, wants a single coordinated plan, or asks which skills to run and in what order.
1 skill
Inventory Intel GPUs (Arc, Arc Pro) on a Linux host. Detect devices, collect driver, firmware and component-health evidence, list processes using each XPU, and read live utilisation.
1 skill
Detect whether a Hugging Face model is text generation, text encoder, seq2seq, masked LM, vision classification, vision-language (CLIP), audio encoder, audio seq2seq, multimodal VL, diffusion, time-series, or a reward model before loading it on Intel XPU. Pairs with torch-xpu-run and vllm-xpu-run so the agent picks the right AutoModel class and input kwargs and avoids wrong-input failures after a 20-second load.
1 skill
Execute a single-target CUDA-to-XPU port of a PyTorch Python repo with libcst-based scan, mechanical rewrite, optional generic availability checks, 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 ./my-model 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.
1 skill
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.
1 skill
Run a read-only go/no-go preflight before any Intel GPU/XPU skillpack work. Checks the GPU driver binding, /dev/dri permissions, render/video groups, Docker, /dev/shm, disk, proxy, and optional container-level XPU visibility; captures health and kernel-log evidence. 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.
1 skill
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.
1 skill