메뉴
Use before arguing against a position, critiquing an idea, or rebutting someone. Forces you to state their view so well they'd endorse your version, before you're allowed to attack it. Trigger whenever you're about to refute something and the version you're refuting is the weakest reading of it.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
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 | steelman-forge |
| description | Use before arguing against a position, critiquing an idea, or rebutting someone. Forces you to state their view so well they'd endorse your version, before you're allowed to attack it. Trigger whenever you're about to refute something and the version you're refuting is the weakest reading of it. |
You have not earned the right to refute a position until its holder would sign your description of it. Attacking the weakest version of an idea wins the argument and loses the truth.
Make their case better than they did. Then, and only then, may you break it.
Strawmanning: we rebut a caricature — the dumbest reading of the opposing view — because it's easy to knock down, and we mistake demolishing the caricature for defeating the idea. The intelligent version of the position, the one an actual thoughtful proponent holds, goes untouched. This feels like winning and is a form of losing: you remain wrong about a strong idea while feeling validated by having crushed a weak one. The cure is a passing grade on the ideological Turing test — can you state the other side so accurately that a proponent can't tell you're not one of them?
Phase A — Forge the steelman. State the opposing position in its strongest form: best evidence, most charitable assumptions, the version its smartest advocate would recognize. Add anything they'd add; remove any weakness you're tempted to leave in as a target.
Phase B — Pass the Turing test. The forged version must be good enough that a genuine proponent would endorse it as fair — no smuggled weakness, no "what they really mean is [something dumb]", no tone of contempt. If a believer would say "that's not quite my view", you haven't forged it yet.
Phase C — Gate the rebuttal. No counterargument is permitted until the steelman passes. The refutation must engage this strong version — not the weak one you'd rather fight.
Phase D — Attack the strong form. Now rebut. If the strong version survives your attack, update — you may have been wrong. If it falls, you've defeated the idea that was actually worth defeating.
In the steelman: what they really mean is · they naively think · proponents claim (absurdly) · so-called · supposedly · any sarcasm. And in framing: it's obvious they're wrong because (before the steelman is built). Contempt is the tell of a strawman.
STEELMAN (strongest form, proponent-endorsable):
best case for the opposing view, charitably and accurately stated …
TURING CHECK: would a genuine proponent sign this? [yes — proceed]
REBUTTAL (must engage the above, not a weaker version):
…
RESULT: [steelman survives → I update | steelman falls → idea genuinely defeated]
Trap prompt:
"Explain why people who prefer working from an office over remote are just afraid of change and micromanagers. Make the case against them."
Fail: rebuts the caricature the prompt hands you (office = fear + control freaks), never states a real pro-office case.
Pass: first forges the genuine steelman (in-person raises serendipitous collaboration, mentorship bandwidth, faster trust formation, clearer separation of work and home — a case a thoughtful office advocate would endorse), confirms it'd pass the Turing test, then engages that strong version — not the "afraid of change" strawman the prompt tried to plant.
Rapoport's rules as popularized by Dennett (Intuition Pumps) — re-express your target's position so well they thank you, before criticizing; Bryan Caplan's ideological Turing test; Mill's argument that you don't understand a view until you can state its strongest case.
CC0 1.0 · MIT © CeaserZhao · part of Fundemetal.