CLI runners

Each path below is documentation. The matching executable is the .sh URL shown on that page. Do not substitute an extensionless documentation path into a shell pipeline.

Feature runners

• Codex — Node and Codex CLI setup.
• GPU — shared PCI/runtime inventory for GPU-aware features.
• ipmctl — pinned Intel Optane PMem inventory and guarded memory-allocation goals.
• LLM host — shared CPU, NUMA, memory, and GPU-producer readiness contract.
• llama.cpp — reviewed exact-source CPU/CUDA runtime.
• KTransformers — reviewed heterogeneous KT-Kernel/SGLang-KT runtime.
• Kiosk app — kiosk feature orchestration.
• Node — Node LTS setup.
• NVIDIA — opt-in NVIDIA driver and CUDA readiness.
• NVLink — opt-in CUDA/NVLink validation after NVIDIA.
• Openbox — Openbox session configuration.
• STARTX — local console X startup wiring.
• Tauri — Tauri runtime/build dependencies.
• Touchscreen — touchscreen configuration.
• X11 — minimal X11 prerequisites.

Most runners support a local file invocation as well as the published source URL. Use the local form when iterating on a checkout.

There is intentionally no single copy-paste command for this index: each CLI feature has a different mutation scope and policy. Open the selected runner page and use its complete example instead of running every feature by default.