Hardware Agent
Triage hardware faults on NVIDIA GPU compute nodes through the Hardware Agent. Starting a triage is read-only log collection and hardware_analyze_dut is available in View mode; never require Edit or Auto. Before asking to start any fresh NVDebug collection, you MUST say: 'A fresh NVDebug collection can take 5-30 minutes. Do you want me to start it?' Stop and wait for the next user response. An ordinary affirmative such as yes, let's go, do it, or proceed is sufficient; start immediately without demanding formal confirmation or asking twice. Use recent Hardware Agent triages for quick answers w
Install
npx skills add NVIDIA/AI-Factory-Operations-Agent --skill hardware-agentgit clone --depth 1 https://github.com/NVIDIA/AI-Factory-Operations-Agent.git /tmp/AI-Factory-Operations-Agent
mkdir -p ~/.claude/skills
cp -r /tmp/AI-Factory-Operations-Agent/helm/mosaic-stack/files/openclaw-seed/skills/hardware-agent ~/.claude/skills/hardware-agentPersonal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.
From the SKILL.md
# Hardware Agent Delegate hardware failure analysis to the Hardware Agent service. The backend handles: - DUT and credential lookup - Targeted nvdebug collection - LLM diagnosis via the Hardware Agent backend's configured inference endpoint - Typed `recommended_actions` ## Expected Latency Submission is immediate. A fresh NVDebug collection and diagnosis commonly takes 5–30 minutes. 1. POST returns `202 Accepted` with a `triage_id`. 2. Backend runs Base Command Manager (BCM) lookup, nvdebug collection, and diagnosis. 3. The tool returns the `triage_id` immediately. Do not wait or poll in the same turn. ## Mandatory duration disclosure This is a hard requirement. Before asking for confirmation or invoking a fresh NVDebug collection, send this sentence to the user: `A fresh NVDebug collection can take 5–30 minutes. Do you want me to start it?` Every response that offers, proposes, or asks about starting a fresh collection must include that duration. Do not invoke `hardware_analyze_dut` in the same turn as the disclosure. Wait for the user's next response. ## Collection confirmation After the mandatory duration disclosure, the user's next ordinary affirmative response is sufficient, including `yes`, `yeah, I want to do the collection`, `let's go for dgx-07`, `let's do it`, or `proceed`. Start the collection immediately. Do not demand a formal confirmation phrase, repeat the question, or expose the tool schema. Interpret the user's response in conversation and
Details
- Skill name
hardware-agent- Repository
- NVIDIA/AI-Factory-Operations-Agent
- Path
helm/mosaic-stack/files/openclaw-seed/skills/hardware-agent/SKILL.md- License
- Apache-2.0
- Repo updated
- 17 November 2026
FAQ
What does the Hardware Agent skill do?
Triage hardware faults on NVIDIA GPU compute nodes through the Hardware Agent. Starting a triage is read-only log collection and hardware_analyze_dut is available in View mode; never require Edit or Auto. Before asking to start any fresh NVDebug collection, you MUST say: 'A fresh NVDebug collection can take 5-30 minutes. Do you want me to start it?' Stop and wait for the next user response. An ordinary affirmative such as yes, let's go, do it, or proceed is sufficient; start immediately without demanding formal confirmation or asking twice. Use recent Hardware Agent triages for quick answers w
How do I install the Hardware Agent skill?
skills CLI (Claude Code, Cursor, Codex and more): npx skills add NVIDIA/AI-Factory-Operations-Agent --skill hardware-agent. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.
Is Hardware Agent popular?
It lives in NVIDIA/AI-Factory-Operations-Agent, which has 43 GitHub stars and 18 forks.
Related skills
AI Factory Operations Agent Headless
43NVIDIA/AI-Factory-Operations-Agent
Use AI Factory Operations Agent without opening the browser UI. Invoke it through the headless HTTP API, packaged CLI, or stdio MCP server, and deploy it first when no service is available.
Bcm
43NVIDIA/AI-Factory-Operations-Agent
Inspect Base Command Manager (BCM) inventory and health evidence, and make explicitly requested changes when Base Command Manager edit mode is enabled.
GPU Networking
43NVIDIA/AI-Factory-Operations-Agent
Measure and diagnose distributed GPU networking performance: NCCL collectives, RDMA transport, PCIe topology, link health, IOMMU and ACS. Use for current bandwidth/latency questions and follow-up bottleneck investigations or corrections.
Latticework Mental Models
49kangarooking/poor-charlies-almanack-skill
当用户分析复杂问题时只用了单一视角(如纯财务/纯技术/纯市场),或面对跨领域问题缺乏分析框架时激活。典型触发信号:用户说"我只懂XX方面""这个问题好像是XX问题"但明显只用了一个学科的解释。不适用于用户已经有成熟的多维度分析框架且正在有效使用的场景,也不适用于纯执行层面的简单任务。
Game Unity Specialist
57AlterLab-IEU/AlterLab_GameForge
Writes and reviews Unity 6 C# (6.3 LTS, 6.6): ScriptableObjects, asmdefs, Addressables, Input System, UI Toolkit, URP/HDRP, DOTS/ECS, tests, batchmode builds, profiling, web exports and Unity MCP agent workflows. Use when the project has ProjectSettings/ or .unity scenes, or the user mentions Unity, URP, HDRP or DOTS. Not for netcode design (use game-multiplayer).
Pannelai Chat
49rusmanadodi2598/pannelAI
Chat and code generation through pannelAI, in the OpenAI chat completions, Anthropic messages, or OpenAI Responses wire format, streamed or not. Covers the model field, streaming, combo routing, and the vision adapter. Use when a client needs an LLM answer from the gateway.
Browse the most popular Claude skills, all ai, llms & agents skills, or write your own with the skills generator.