菜单
Build local evidence Q&A artifacts with source finding IDs and no external LLM dependency.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Analyze slow AX frontends, problematic users/clients, broad inventory queries, and machine-wide impact without installing client parsers.
Build or run Autonomous Ops workflows for AX Performance Advisor, including investigation queues, evidence acquisition plans, follow-up questions, change drafts, validation planning, and readiness gates.
Generate AI learning and decision-quality artifacts such as recommendation memory, similarity search, acceptance simulation, narrative variants, anomaly explanations, and action confidence tuning.
Generate autonomous AI/USP investigation artifacts including evidence scout, investigation tree, root-cause debate, recommendation quality gate, KPI storyboard, and anonymized pattern library.
Generate guarded admin execution previews with policy gates, confirmation tokens, audit records, and rollback/validation metadata.
Generate advanced operational USP outputs including SLO burn rate, maintenance sequencing, cost of delay, release gates, retention candidates, known issue matching, and executive briefings.
| name | local-rag-evidence-qa |
| description | Build local evidence Q&A artifacts with source finding IDs and no external LLM dependency. |
Run scripts/rag_qa.py with --evidence, --output, and optional --question. Answers must cite source finding IDs and avoid unsupported claims.
