iraop
Skill37 starsChanged 10 days agoApache-2.0View on GitHub
Research NVIDIA DGX, SuperPOD, GB200, NVLink, BlueField, Base Command Manager (BCM), and on-prem infrastructure documentation through the IRA/Sequoia retrieval agent. Use when the user asks to research docs, deployment policy, operational guidance, or newly uploaded knowledge-base documents.
In this repository8 other files it gives its agents.
- helm/mosaic-stack/files/openclaw-seed/workspace/AGENTS.mdAGENTS.md
- ai-factory-operations-agent-headlessSkill
- bcmSkill
- gpu-networkingSkill
- grafanaSkill
- hardware-agentSkill
- observabilitySkill
- slurmSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 0 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 21
Use the iraop_query tool for documentation-backed research.
- Very specific: line 29
Default to depth: "quick" unless the user explicitly asks for deep research or the question needs a multi-step plan grounded in …
- Very specific: line 31
If the user references uploaded knowledge, first use iraop_list_collections or pass the likely collection name to iraop_query when known.
- Very specific: line 35
When the user asks for a safe Kubernetes configuration command, put the command in a fenced code block and use kubectl apply -f - <<EOF …
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
<!--
SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
SPDX-License-Identifier: Apache-2.0
-->iraop — IRA Documentation Research
Use the iraop_query tool for documentation-backed research.
For an operational investigation, first gather the workload's selected resources, observed symptom and constraints. Include those facts in the research question and ask for applicable diagnostic checks and their prerequisites. A bare health-check label does not identify a cause. Keep unrelated findings out of the remedy question unless live evidence connects them to the affected path.
Research returns documentation, not observations of the running cluster. Before acting on a recommendation, verify its prerequisite on the affected resource with a live read. Keep an explicit distinction between a documented possibility, an observed condition and a cause demonstrated by a controlled comparison. If a change does not improve the measured symptom, it has not resolved the task. Never infer normal performance from the absence of a successful remedy; that claim requires a matching benchmark reference.
Reuse an existing research answer until new live evidence creates a different question. Rephrasing the same question without new evidence does not advance the investigation.
Default to depth: "quick" unless the user explicitly asks for deep research or the question needs a multi-step plan grounded in documentation. Use depth: "deep" for deployment-policy recommendations, operational plans, or questions that should synthesize multiple documents.
If the user references uploaded knowledge, first use iraop_list_collections or pass the likely collection name to iraop_query when known.
Return practical SRE-oriented output: findings, caveats, and exact safe commands when the user asks for commands. Do not create GPU pods unless the user explicitly requests a real rollout.
When the user asks for a safe Kubernetes configuration command, put the command in a fenced code block and use kubectl apply -f - <<EOF with inert resources such as namespaces or ConfigMaps. Never output kubectl create for this workflow, including namespace or ConfigMap examples; use apply-style YAML only.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

