Agent Skills
› mohitagw15856/pm-claude-skills
› what-to-ask
what-to-ask
GitHub在签署、购买或达成协议前，根据具体场景生成五个最具杠杆效应的关键问题。每个问题附带重要性说明、糟糕回答示例及书面确认要求，并自动路由至相关深度解析工具，帮助用户规避风险。
Trigger Scenarios
即将签署合同或协议
准备购买大件商品或服务
询问签约前应问的问题
涉及房东、经销商、承包商或HR的谈判
Install
npx skills add mohitagw15856/pm-claude-skills --skill what-to-ask -g -y
SKILL.md
Frontmatter
{
"name": "what-to-ask",
"description": "Get the five questions that matter before you sign, buy, or agree to anything — the front door to the decoder family, routed by situation. Use when asked what should I ask before signing this, I'm about to buy X what do I check, what questions for the landlord\/dealer\/contractor\/HR, or what am I forgetting. Produces the five highest-leverage questions for the specific situation with why each matters and what a bad answer sounds like, plus the pointer to the full decoder when one exists."
}
What To Ask Skill
Every consequential signature has five questions that would have changed everything — asked the day before instead of discovered the year after. This skill is the library's front door: name the situation ("signing a lease," "buying a used car," "joining a startup," "hiring a contractor"), get the five questions that carry the most leverage for that situation, each with why it matters and what a bad answer sounds like. When a full decoder exists for the document, this skill hands off to it; when none does, five good questions still beat walking in empty.
What This Skill Produces
- The five questions — highest-leverage first, phrased ready to say out loud
- Per question: why it matters (the failure it prevents) and what a bad answer sounds like (the tell to listen for)
- The get-it-in-writing flags — which answers must survive on paper
- The handoff — the full decoder/calculator/simulator for this situation, when the library has one
Required Inputs
Ask for these if not provided:
- The situation — what's being signed/bought/agreed, with whom, and when (tomorrow changes the advice from "research" to "triage")
- The stakes and the worry — money involved, and the thing they're privately nervous about (the fifth question is usually theirs)
- What's already known — documents in hand get routed to their decoder; verbal-only situations get the questions that force things onto paper
Framework: The Question-Selection Rules
- Leverage order, not checklist order: the five are chosen by expected-cost-prevented, not thoroughness — the question that voids the deal outranks ten that adjust it. A list of twenty is a list of zero; five is the format because it forces the ranking.
- Every question gets its bad-answer tell: "What happens if I need to exit early?" matters less than knowing that "oh, we're flexible, we'll work with you" is the bad answer — vagueness where numbers should be is the universal tell, and each question names its specific version.
- The writing question is always in the five for anything with money attached: some form of "can I get that in writing?" — because the answers to questions 1–4 are worth exactly what they're written on.
- Route to depth when it exists: lease → lease-decoder; job offer → benefits-decoder; contractor → home-contractor-quote-decoder; car lot → the-car-dealership to rehearse it first. The five questions are the door; the decoder is the room.
- Unknown situations still get real questions — derived from the universal five axes: exit costs (how do I leave?), change costs (what can they change on me?), the money's full shape (what's the all-in number?), failure modes (what happens when something goes wrong?), and verification (how do I check what you just told me?). Any situation on earth maps onto those.
Output Format
Before you [sign/buy/agree]: [situation]
The Five
1. "[The question, verbatim]" Why: [the failure it prevents] · Bad answer sounds like: "[the tell]" · [📝 get in writing]
[…2–5, same shape…]
If You Only Ask One
[Which, and why it carries the most weight here]
Go Deeper
[The library's full treatment for this situation, linked — or "no decoder yet; the five axes above are the map"]
Quality Checks
- Exactly five questions, leverage-ranked — not a padded checklist
- Every question has a bad-answer tell, specific to it
- The in-writing flag appears wherever money does
- The user's stated private worry became one of the five
- The handoff link appears when the library covers the situation
Anti-Patterns
- Do not exceed five — the ranking discipline is the product; a sixth question weakens the first
- Do not ask questions the user can't act on — every answer must change something
- Do not phrase questions adversarially — the counterpart is a person; the tells do the detecting
- Do not skip the handoff to save face — five questions are the trailer, not the movie
- Do not generate generic questions for a specific situation — "what are the terms?" is what this skill exists to replace
Version History
- 54fad50 Current 2026-07-19 13:47
