Claude Skills by air-gapped

NVIDIA AIPerf — vendor-neutral generative-AI inference benchmarking (genai-perf successor). Covers `aiperf profile` with concurrency / request-rate / fixed-schedule trace replay / user-centric / multi-run confidence, 15 endpoint types (chat, completions, embeddings, rankings, responses, image-gen, video-gen, NIM, HF-TGI, template, etc.), 6 custom dataset formats (single_turn, multi_turn, mooncake_trace, bailian_trace, burst_gpt_trace, random_pool), 40+ public datasets, goodput SLOs, GPU + Pro...

Run and debug `dellemc.openmanage` Ansible playbooks against Dell PowerEdge **iDRAC 9** (14G–16G) and **iDRAC 10** (17G — R670, R770, R870, R970, XE9780, XE9785). Covers the iDRAC 10 / iDRAC 9 ≥ 7.30.10.50 `BasicAuthState: Unadvertised` default that silently 401s `ansible.builtin.uri` (Dell KB 000437501), the `idrac_session` + `x_auth_token` lifecycle with `block:/always:`, `force_basic_auth: true` fallback for raw Redfish, OMSDK modules (`idrac_firmware`, `idrac_server_config_profile`) that ...

Author and maintain Argo CD `Application` and `ApplicationSet` manifests as a GitOps consumer (publisher), targeting Argo CD v3.3 / v3.4 (May 2026). Covers source types (Helm, Kustomize, OCI, multi-source, plugin), sync policies + options + waves + hooks, ApplicationSet generators (List, Cluster, Git, Matrix, Merge, SCMProvider, PullRequest, Plugin, ClusterDecisionResource), Progressive Sync (Beta), Source Hydrator (still Alpha), AppProjects, RBAC, sync impersonation (`destinationServiceAccou...

Karpathy-pattern autoresearch — autonomous hill-climbing over a measurable metric, deep multi-agent research, or research-then-optimize. Three modes: Optimize (keep/discard ratchet), Research (STORM multi-perspective), Improve.

BAML (Boundary ML) expert for projects defining LLM calls as typed functions in .baml files with a generated Python client. Use whenever the repo contains baml_src/, baml_client/, baml-cli commands, or imports from baml_py / baml_client. Covers .baml syntax (function, class, enum, client, test, retry_policy, attributes), Python integration (baml_client sync/async, streaming, ClientRegistry, Collector, TypeBuilder), Schema-Aligned Parsing, ctx.output_format, @@assert / @@check tests, @stream.d...

Audit AND tune Linux/GPU inference hosts — read-only host snapshot

Author and maintain Helm charts: create chart, write templates, values.yaml patterns, _helpers.tpl, Chart.yaml, values.schema.json, helm-docs, library charts. Helm 4 (SSA, WASM, OCI digest). Chart CI/CD (lint, helm-unittest, chart-testing, kubeconform, OCI push). OpenShift compatibility (SCCs, Routes, UIDs, adaptSecurityContext). Chart security (SecurityContext, RBAC, NetworkPolicy, cosign signing, image digest pinning). CRD management, ServiceMonitor templates, HPA, persistence, resource pre...

Author, read, and debug Jinja2 templates across the three places Jinja lives in 2026 — HuggingFace `chat_template.jinja` (rendered by `apply_chat_template` for vLLM / sglang), Ansible playbooks + `.j2` files, and Jinja-adjacent Kubernetes workflows (`values.yaml.j2`, `kubernetes.core.k8s + template`, Helm post-renderers). Companion to the `helm` skill — Helm charts are Go `text/template` + Sprig, not Jinja, and this skill makes that disambiguation explicit.

Configure, operate, and master KEDA (Kubernetes Event-driven Autoscaling) — ScaledObject, ScaledJob, TriggerAuthentication CRDs, 70+ scalers, HPA behavior tuning, scale-to-zero, the KEDA HTTP Add-on, production hardening, multi-trigger semantics, scalingModifiers formulas, GitOps integration, and troubleshooting stuck scalers. Covers the common traps (cooldownPeriod only applies to N→0, CPU/memory cannot drive scale-to-zero alone, activationThreshold vs threshold, multi-trigger max-of semanti...

Operate, configure, deploy, secure, and integrate with Keycloak (open-source IAM) — the modern Quarkus distribution (24.x–26.6.x), the Keycloak Operator with `Keycloak` and `KeycloakRealmImport` CRDs, and realm/client/identity-provider configuration.

LMCache multiprocess (MP) mode — standalone LMCache server in its own pod/process that vLLM connects to over ZMQ. Provides process isolation, no GIL contention on the inference path, one cache shared by multiple vLLM pods on the same node, and CPU-memory scaling independent of GPU memory. Covers the `LMCacheMPConnector` path (the new direction; `LMCacheConnectorV1` in-process path still works but is being upstaged), DaemonSet+Deployment K8s pattern, L1 (CPU DRAM) + L2 (NIXL POSIX / GDS / HF3F...

Makefile best practices, patterns, and templates for GNU Make 4.x — dependency graphs, task-runner workflows, parallel-safe recipes, self-documenting help targets, and language-specific patterns (Go, Python, Node, Docker, Helm, POSIX).

Bring up NVIDIA HGX/DGX datacenter GPU hosts on Ubuntu 24.04 LTS — air-gapped or connected, Secure Boot enabled. Covers B300/B200/H100/A100/L40S/L4 driver+fabricmanager+NVLSM+DOCA-OFED install order and exact package set from NVIDIA CUDA repo + DOCA repo. Triggers on B300/B200/HGX/DGX install, "fabricmanager won't start", "system not yet initialized" / cudaErrorSystemNotReady, NVLSM missing, ib_umad not loading, DOCA-OFED before NVIDIA driver, nvidia-driver-pinning-XXX, nvlink5-XXX, nvidia-op...

NVIDIA Inference Xfer Library (NIXL) operator + developer reference. Point-to-point KV-cache and tensor transport for distributed inference (Dynamo, vLLM, SGLang). Covers the C++/Python/Rust agent API, all 13 backend plugins (UCX, GDS, GDS_MT, libfabric, mooncake, posix, hf3fs, obj/S3, azure_blob, gusli, uccl, gpunetio/DOCA, telemetry), build paths (pip nixl-cu12/cu13, meson+ninja from source), ETCD vs side-channel metadata, telemetry (Prometheus + cyclic shared-memory), NIXL-EP elastic MoE d...

Wire HuggingFace embedding + reranker models (BGE-M3, BGE-Reranker-v2-m3, etc.) into Open WebUI's RAG pipeline via LiteLLM as a proxy in front of HuggingFace Text Embeddings Inference (TEI). Covers the exact wire shapes Open WebUI sends (URL auto-append on embed but NOT rerank; payload + response shapes for both modes), the LiteLLM ↔ TEI gotchas (encoding_format=null trap, HF-driver task_type misdetection, openai-driver vs huggingface-driver tradeoffs), TEI configuration cliffs (max-client-ba...

Deploy Open WebUI multi-pod with WebSockets and Valkey/Redis Sentinel at 1000+ user scale on Kubernetes. Centerpiece is the structural Socket.IO+Redis frame-amplification bug (#23733) that cripples multi-pod streaming, and the maintainer-endorsed mitigation (`CHAT_RESPONSE_STREAM_DELTA_CHUNK_SIZE`). Covers all multi-pod env vars, the custom-model-icon perf history (base64-in-/api/models, fixed late 2025–Apr 2026), the official helm chart's gaps (bundled Redis is unsuitable for production; no ...

Package applications for OpenShift deployment: container images (UBI, arbitrary UID, multi-stage builds), packaging formats (Helm, Kustomize, Operators, OLM v1), CI/CD (Tekton, ArgoCD, Shipwright, Conforma), security (SCC, PSA, supply chain, image signing, secrets), operations (Routes, probes, scaling, monitoring, storage), disconnected/air-gapped patterns, and critical gotchas. Covers OCP 4.14-4.21. NOT for cluster installation or infrastructure management.

Query Prometheus and Grafana Mimir, write and debug PromQL, and build or fix Grafana dashboards — for agents solving problems from metrics. Covers the Prometheus HTTP API (`/api/v1/query`, `query_range`, `series`, `labels`, `metadata`), Mimir multi-tenancy (`X-Scope-OrgID`, federation `a|b|c`, per-tenant 422/429 limits), the PromQL surface (selectors, rate family, classic + native histograms, `histogram_quantile`, vector matching `on()`/`group_left`, recording rules), Grafana dashboard JSON (...

SGLang HiCache (hierarchical KV cache) — three-tier prefix cache: GPU HBM (L1) → pinned host DRAM (L2) → distributed L3 (Mooncake / 3FS / NIXL / AIBrix / EIC / SiMM / file / LMCache). Covers `--enable-hierarchical-cache`, all `--hicache-*` flags, write policies, page_first* layouts, prefetch policy (best_effort / wait_complete / timeout), per-rank sizing, MHA / MLA / DSA / Mamba / SWA support matrix (hybrid attention still landing in v0.5.11), runtime attach/detach HTTP admin, and auto-rewrit...

SGLang Model Gateway (`sgl-model-gateway`, formerly `sgl-router`) — Rust router fronting vLLM and SGLang inference workers on Kubernetes. Covers first-class vLLM gRPC backend plus HTTP transparent-proxy for vanilla vLLM, eight load-balancing policies, tokenizer-format dispatch (`tokenizer.json` HF-fast vs `tiktoken.model` BPE — including when neither is required because `cache_aware` is text-based), air-gapped recipe (gateway ignores `HF_ENDPOINT`, mount tokenizer files on PVC only when actua...

Autoresearch loop for Claude Code skills — greedy keep/discard hill climbing on a 10-dimension quality rubric, with blind subagent validation for self-scoring bias, plus a `freshen` mode that probes external references (release notes, docs, deprecation signals) and applies verified updates, plus a `trigger` mode that measures and tunes the skill's frontmatter description until it reliably fires when it should and stays silent when it shouldn't (60/40 train/test split, 3 runs/query, blinded te...

Preflight reference for HuggingFace snapshots — what vLLM, sglang, and transformers.generate see at runtime. Covers config-file precedence (tokenizer.json, tokenizer_config.json, generation_config.json, chat_template.jinja), transformers v5 tokenizer-class taxonomy (TokenizersBackend, PythonBackend, MistralCommonBackend, TikTokenTokenizer), special-token discovery (all_special_ids, added_tokens_decoder, extra_special_tokens, backend_tokenizer.get_added_tokens_decoder), chat-template Jinja con...

Run production vLLM benchmarks — `vllm bench` (serve, throughput, latency, sweep, startup, mm-processor), request-rate vs max-concurrency semantics, TTFT/TPOT/ITL/E2EL percentiles, goodput SLO measurement, prefix-cache workloads, air-gapped operation (HF_ENDPOINT, ModelScope, hf-mirror, offline cache). Methodology split — SLO health checks vs A/B change sweeps — plus pitfalls that produce misleading numbers (no warmup, wrong tokenizer, random-as-prod, `--request-rate inf` alone).

vLLM tiered KV cache configuration for production H100/H200 clusters. Native CPU offload, LMCache (CPU+NVMe+GDS), NixlConnector (disaggregated prefill), MooncakeConnector (RDMA), MultiConnector composition. Version gates, sizing math (flag total across TP, not per-GPU — opposite of SGLang), KV-vs-weights offload distinction operators most often get wrong.

vLLM chat-template (prompt-side Jinja) operator reference. Template resolution precedence (`--chat-template` → AutoProcessor → tokenizer default → bundled fallback), `chat_template_kwargs` allowlist silently dropping `add_generation_prompt`/`enable_thinking`/custom kwargs (PR 27622 fix), 27 shipped `tool_chat_template_*.jinja` files, known template-layer bugs for Qwen3/Qwen3-Coder, DeepSeek-R1/V3/V3.1/V3.2, GPT-OSS, Kimi-K2, Llama-4, Mistral (HF vs mistral mode), Gemma-3/4, Phi-4, GLM. Prompt...

Configure vLLM completely — YAML config file format, CLI arg precedence, full VLLM_*/HF_*/TRANSFORMERS_* env-var catalog, end-to-end recipe for air-gapped environments (internal HF mirrors, hf-mirror.com, ModelScope, HF_HUB_OFFLINE with pre-seeded cache, gated models offline, trust_remote_code supply-chain implications). VLLM_HOST_IP vs API-host confusion, Kubernetes-service-named-`vllm` env-var poisoning, usage-stats triple opt-out, YAML precedence surprises.

Use this skill when authoring, reviewing, or fixing a vLLM Kubernetes manifest, Docker/Podman pod, or OpenShift ServingRuntime — even when the user does not say "vllm". Triggers on: lab cluster performance practices, cache mount + survival across pod restarts (/root/.cache, VLLM_CACHE_ROOT, TORCHINDUCTOR_CACHE_DIR, TRITON_CACHE_DIR, "do we have caches saved"), HF_TOKEN secret in pod env, liveness + readiness probe tuning (initialDelaySeconds, failureThreshold, "pod takes 12 min to boot"), ser...

Operating-point reference for serving Gemma 4 31B on vLLM — TP sizing, max_model_len, max_num_seqs, gpu_memory_utilization, kv_cache_dtype, EAGLE3 spec-dec, chat_template choice.

vLLM non-chat inference surfaces — text embeddings (`/v1/embeddings`, `/v2/embed`), reranking/scoring (`/rerank`, `/score`), speech-to-text (`/v1/audio/transcriptions`, `/v1/audio/translations`), document OCR via VLMs. Covers 2026 `--runner pooling` (replacing `--task embed`), v0.20 deprecations (`score`→`classify`, multitask pooling, `encode`→`token_embed`+`token_classify`), Matryoshka/MRL, ColBERT/ColPali/ColQwen late-interaction MaxSim, Cohere v2 `/v2/embed`, Jina v3/v4/v5 quirks, cross-en...

NVIDIA AI-hardware + vLLM-platform reference covering Hopper (H100/H200), Blackwell (B100/B200/B300) and Blackwell Ultra, Grace-Blackwell superchips and NVL72 racks (GB200, GB300), Vera Rubin (R100/R300) with VR200 NVL144 and Kyber NVL576, Dell PowerEdge XE family and IR5000/IR7000/IR9048 racks. Per-SKU HBM, FP4/FP8/FP16 TFLOPs, NVLink5, TDP, rack power/cooling (135 kW GB300, 180-220 kW NVL144, 600 kW Kyber), DLC vs RDHx, 800 VDC HVDC. Memory-wall roofline, HBM3E→HBM4 supply 2026. vLLM attent...

Observe production vLLM — `/metrics` Prometheus surface (V1 engine), SLO-driven alerting on TTFT/ITL/queue/KV/preemption/aborts/corrupted-logits, shipping Grafana dashboards in `examples/observability/`, OTLP tracing with `--otlp-traces-endpoint` and `--collect-detailed-traces={model,worker,all}`, diagnostic rules to triage from /metrics alone — queue-grows + TPOT-stable means capacity, queue-stable + TPOT-grows means context/model, DCGM `SM_OCCUPANCY` is the real GPU-saturation signal not `G...

vLLM-Omni output-side multimodal generation — image (FLUX.1/2, Qwen-Image, GLM-Image, BAGEL, SD3.5, HunyuanImage-3.0), video (Wan2.1/2.2, LTX-2, HunyuanVideo-1.5), TTS (Qwen3-TTS, CosyVoice3, Voxtral-TTS), any-to-any omni (Qwen3-Omni, Qwen2.5-Omni, MiMo-Audio) via `vllm serve --omni`. Stage-based disaggregation (OmniConnector + Mooncake + RDMA), `/v1/images/generations`, async+sync `/v1/videos`, `/v1/audio/speech` with voice-upload, PCM16 WebSocket `/v1/realtime`, Ulysses/Ring SP + CFG-parall...

vLLM performance-tuning operator reference — tuning workflow (baseline → bottleneck → knob → re-bench), fused-MoE kernel autotune (`benchmark_moe.py` generates `E=N,N=M,device_name=X.json` configs), DeepEP all-to-all + expert parallelism + EPLB, CUDA graph modes (FULL_AND_PIECEWISE default), torch.compile AOT + compile cache, scheduler knobs (`--max-num-batched-tokens`, `--max-num-seqs`, `--async-scheduling`), TP/EP/DP/PP decision tree, NCCL/DCGM on H100/H200/B200/GB200, PD disaggregation (Ni...

vLLM datacenter-GPU quantization — picking, configuring, troubleshooting NVFP4, FP8, MXFP4, MXFP8, AWQ, GPTQ, INT8, compressed-tensors, modelopt, quark on H100/H200/B200/B300/GB200/GB300. 29 `--quantization` flag values, KV-cache dtypes (fp8_e4m3, nvfp4, per-token-head, turboquant), MoE backend selection (CUTLASS, TRTLLM, FlashInfer, DeepGEMM, Marlin, Qutlass), producing checkpoints with llm-compressor and NVIDIA ModelOpt (NVFP4_DEFAULT_CFG, FP8_DEFAULT_CFG, W4A16, SmoothQuant+GPTQ), online q...

Survey an RKE2 community cluster against an embedded compatibility registry of 18 stack components and produce a verdict for upgrade-readiness, drift-review, and version-skew questions. Components: RKE2, Rancher, Harvester, Cilium, cert-manager, Kyverno, KEDA, Argo CD, Harbor, Traefik, Rook, Ceph, OpenEBS, GitLab, ECK, Zalando postgres-operator, Grafana Mimir, NVIDIA GPU Operator. Works air-gapped — all per-component compatibility data lives in `references/compat/` and is maintained out-of-ba...

vLLM reasoning-parser operator + developer reference. `--reasoning-parser` CLI wiring, `ReasoningParser` contract (non-streaming `extract_reasoning` + per-delta `extract_reasoning_streaming`), `is_reasoning_end` xgrammar gating, `--structured-outputs-config.enable_in_reasoning` bypass, 22 built-in parsers with per-model quirks, 15 production pitfalls, authoring custom parsers via `@ReasoningParserManager.register_module` or plugin.

Pick, configure, tune, monitor vLLM speculative decoding in production. Eleven SpeculativeMethod options (ngram, ngram_gpu, medusa, mlp_speculator, draft_model, suffix, eagle, eagle3, dflash, mtp, extract_hidden_states), `--speculative-config` JSON schema, which methods pair with which target model family, Prometheus acceptance metric surface, version gates (v0.11.1 EAGLE-3 preamble fix, v0.16 parallel drafting, v0.18 ngram_gpu, v0.19 dflash and zero-bubble), composability with chunked prefil...

vLLM tool-calling operator reference — picking `--tool-call-parser` per model family, writing custom parsers via `--tool-parser-plugin`, navigating vLLM source + GitHub tracker to debug any specific tool-call question. Pointer map, not source paraphrase. All 36+ built-in parsers (JSON-sentinel, pythonic, XML, harmony grammars), CLI contract (`--enable-auto-tool-choice`, `--reasoning-parser` pairing, chat-template selection), framework contract (`prev_tool_call_arr` + `streamed_args_for_tool` ...