inference-host plugin
Inference host bring-up + tuning suite — Linux/GPU bare-metal host work that sits beneath any inference framework (vLLM, sglang, TensorRT-LLM). Two phases. Bring-up (nvidia-datacenter-bringup): Ubuntu 24.04 LTS from clean OS to gpu-operator cuda-validator passing — B300/B200/H100/A100/L40S/L4 driver + fabricmanager + NVLSM + DOCA-OFED install order, NVIDIA CUDA repo + DOCA repo + air-gap mirror, MOK + DKMS sign-on-build under Secure Boot, Dell PowerEdge XE9780/XE9785 baseboard firmware via iDRAC Redfish DellOemChassis.ExtendedReset, gpu-operator pre-installed-driver-mode integration. Tuning (gpu-host-tuning): read-only snapshot of CPU power state, C-states, NUMA topology, PCIe link state, GPU settings, kernel boot params, sysctl, ulimits, IRQ affinity, container runtime; optional pinned-host↔GPU memcpy bench (torch + numactl); per-lever cheat-sheets to flip settings (governor, EPP, cpuidle, persistence, ECC, hugepages, intel_iommu, NCCL env, tuned-adm profiles, BIOS guidance for Dell X
How to install the inference-host plugin
Run both commands inside Claude Code. The first adds the marketplace once; the second installs the plugin.
/plugin marketplace add air-gapped/skills
/plugin install inference-host@air-gapped-marketplaceWhat's inside inference-host
2 parts read from the plugin's own files.
Skills 2
Claude loads it when the task matches- gpu-host-tuning
Audits and tunes Linux host configurations to optimize performance for GPU-based inference frameworks.
- nvidia-datacenter-bringup
Brings up NVIDIA HGX/DGX GPU hosts on Ubuntu 24.04 LTS, including driver, fabricmanager, and DOCA-OFED installation for air-gapped or connected environments.
Using inference-host in Claude Code
Skills need no command. Claude reads them when your request matches what they describe.
Update, turn off or remove
- Update the plugin (shell)
- claude plugin update inference-host@air-gapped-marketplace
- Refresh the marketplace
- /plugin marketplace update air-gapped-marketplace
- Turn it off for now
- /plugin disable inference-host@air-gapped-marketplace
- Remove it
- /plugin uninstall inference-host@air-gapped-marketplace
Skills in inference-host
- gpu-host-tuningair-gapped/skills
Audits and tunes Linux host configurations to optimize performance for GPU-based inference frameworks.
15 files6stars - nvidia-datacenter-bringupair-gapped/skills
Brings up NVIDIA HGX/DGX GPU hosts on Ubuntu 24.04 LTS, including driver, fabricmanager, and DOCA-OFED installation for air-gapped or connected environments.
20 files6stars
inference-host plugin questions
How do I install the inference-host plugin?
In Claude Code run /plugin marketplace add air-gapped/skills, then /plugin install inference-host@air-gapped-marketplace.
What does the inference-host plugin include?
inference-host bundles 2 skills.
How do I update or remove the inference-host plugin?
Update it with claude plugin update inference-host@air-gapped-marketplace in your shell, or from the Installed tab of /plugin. Remove it with /plugin uninstall inference-host@air-gapped-marketplace.
What license is the inference-host plugin under?
MIT, as stated in the plugin's manifest.