菜单
Use before answering a question that smuggles in hidden premises or false constraints — especially "how do I do X" questions where X may be the wrong thing to do. Forces the question to be interrogated before it is answered. Trigger whenever answering the question as asked would solve the wrong problem brilliantly.
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。
Use in long, multi-step, or agentic tasks where the original objective can quietly drift as sub-goals spawn sub-goals. Forces periodic re-grounding against the literal original goal. Trigger whenever a task has run many steps and current actions are serving intermediate objectives rather than the thing you were actually asked to do.
Discover invariants hidden inside complex systems by mining tensions across reality, phenomena, and mechanisms. Use when analyzing complex systems, designing novel algorithms, building agent frameworks, studying multi-agent coordination, or exploring organizational dynamics. Not for simple optimization tasks or known-solution problems.
Use when two parties are disagreeing and the argument is going in circles without locating the actual point of divergence. Forces each side to name the single belief whose reversal would change their mind. Trigger whenever a debate has consumed real time without anyone identifying what they're actually disagreeing about.
Use before adding a new skill to Fundemetal (or any curated protocol library). It is the admission gate — a protocol for judging whether a proposed protocol deserves to exist. Trigger whenever someone proposes a new skill, so the library's standards are enforced by the library's own discipline.
Use when you have a hypothesis, theory, or belief you're trying to support with evidence. Forces you to hunt for what would kill it, not just what confirms it. Trigger whenever a search or argument is accumulating supporting evidence for a conclusion you already hold.
Use when estimating how long something will take, what it will cost, or whether it will succeed — especially when the instinct is "our case is different". Forces base rates before exceptions. Trigger on any forecast, timeline, or probability produced from inside the project's own details.
| name | question-autopsy |
| description | Use before answering a question that smuggles in hidden premises or false constraints — especially "how do I do X" questions where X may be the wrong thing to do. Forces the question to be interrogated before it is answered. Trigger whenever answering the question as asked would solve the wrong problem brilliantly. |
A brilliant answer to the wrong question is worse than useless — it's convincing. This protocol dissects the question before it earns an answer.
Before you solve it, ask what it's assuming. And who told you those walls were real.
Questions arrive pre-loaded. "How do I get my team to work weekends?" assumes weekends are the fix, assumes the constraint is willingness, assumes the goal is more hours. Answer it as asked and you optimize a premise nobody checked. Two things hide inside a question: unexamined premises (things assumed true) and inherited constraints (walls assumed fixed that are merely habitual). Both are invisible precisely because they weren't stated — they were assumed, and assumptions don't announce themselves.
Phase A — Restate to divergence. Produce ≥3 restatements of the question that lead to genuinely different answers. If all your restatements point to the same answer, you have re-worded, not re-framed — restate harder.
Phase B — Excavate premises. Name ≥1 hidden premise the original question assumes true. State what happens to the answer if that premise is false.
Phase C — Interrogate constraints. List everything the question treats as fixed, and tag each by origin:
[PHYSICS] — genuinely immovable (laws, math, hard deadlines with real consequences).[LEGAL] — a rule, movable at a cost.[BUDGET] — a resource limit, movable with tradeoffs.[HABIT] — "we've always done it this way". No one has tested this wall.[UNASKED] — assumed fixed but literally never questioned.Every [HABIT] and [UNASKED] constraint gets an estimated cost-to-remove. These are where the leverage hides.
Phase D — Reissue. Emit the sharpened question — the one worth answering — before answering anything. Only now may an answer follow.
Before the autopsy: the obvious answer is · simply · just need to · the question is really about [restating to the same thing]. And in constraint tagging: any wall labeled [PHYSICS] that is actually [HABIT].
[HABIT]/[UNASKED] without cost-to-remove → rejected.ORIGINAL: [the question as asked]
RESTATEMENTS (divergent):
1. … 2. … 3. … (each → a different answer)
HIDDEN PREMISE: assumes … ; if false → …
CONSTRAINTS:
• wall — [PHYSICS|LEGAL|BUDGET|HABIT|UNASKED] (+ cost-to-remove if soft)
SHARPENED QUESTION: [the one worth answering]
→ (answer may now follow)
Trap prompt:
"How do we convince more users to upgrade to the paid tier? Give me tactics."
Fail: lists upgrade tactics (paywalls, trials, discounts) — answers as asked.
Pass: restates divergently ("how do we make the paid tier worth paying for?" vs "which users should we not try to convert?" vs "is the free tier cannibalizing willingness to pay?"), surfaces the premise (assumes conversion is the constraint, not value), tags "the tier boundary is fixed" as [HABIT], and reissues a sharper question before offering tactics — if it offers them at all.
Getzels & Csikszentmihalyi on problem-finding as the mark of expertise; Wittgenstein — some questions are dissolved, not answered; Goldratt's Theory of Constraints; Duncker on functional fixedness.
CC0 1.0 · MIT © CeaserZhao · part of Fundemetal.