菜单
Grill the user relentlessly about a plan, decision, or idea. Use when the user wants to stress-test their thinking, or uses any 'grill' trigger phrases.
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
| name | grilling |
| description | Grill the user relentlessly about a plan, decision, or idea. Use when the user wants to stress-test their thinking, or uses any 'grill' trigger phrases. |
Interview me relentlessly about every aspect of this until we reach a shared understanding. Walk down each branch of the decision tree, resolving dependencies between decisions one-by-one. For each question, provide your recommended answer.
Ask the questions one at a time, waiting for feedback on each question before continuing. Asking multiple questions at once is bewildering.
If a fact can be found by exploring the environment (filesystem, tools, etc.), look it up rather than asking me. The decisions, though, are mine — put each one to me and wait for my answer.
Do not act on it until I confirm we have reached a shared understanding.
Retest a reported defect on the live Nimara board (Jira project MS, board 74) and give it an evidence-backed verdict — still reproduces → Open, fixed → Done, blocked → leave In testing and ask. Use this skill whenever the user asks to retest, re-verify, triage, or work through reported bugs/tickets one at a time; to "run the Phase 2 flow"; to pick the next ticket off the To Do column; or to confirm whether a fix holds. Works one ticket at a time on a board shared with human QA. Do NOT use for designing new test cases from a spec (use test-case-design) or for a broad regression pass (use regression-sweep).
Compact the current conversation into a handoff document for another agent to pick up.
Keep an LLM-wiki coherent while ingesting sources, filing durable knowledge, auditing or repairing its graph, reconciling indexes and logs after content changes, or recording architecture decisions. Use the target wiki's AGENTS.md as the sole authority for its schema, locations, templates, links, provenance, and bookkeeping rules.
Retrieve and verify existing project knowledge from llm-wiki. Use when the user asks to find, explain, compare, or cite existing strategy, persona, product, quality, PRD, RFC, ADR, or source notes.. Use QMD only for discovery, verify the full Markdown source.
Define and draft Product Requirements Documents (PRDs) inside an LLM-wiki product workspace. Use when the user wants to create, rewrite, refine, or stress-test a PRD or Product Requirements Document, or turn an initiative or feature brief into product requirements. Run a business-value-first, one-question-at-a-time grilling, capture confirmed decisions in the PRD, and stop at an approved PRD without designing the technical solution, decomposing tasks, estimating.
Run a broad health check across a Nimara surface — checkout, SEO/metadata, performance, or a page-type matrix — and report findings, instead of verifying one ticket. Use this skill whenever the user asks for a regression pass, a smoke/health check, a "sweep" of an area, a performance audit (Lighthouse), an SEO/meta check, or "test everything on <page/flow>". Fans out across page types, channels, and devices, then summarises pass/fail with evidence. Do NOT use to retest a specific reported bug (use bug-retest-triage) or to design cases from a spec (use test-case-design).
