메뉴
Use when multiple agents (or people) will weigh in on the same question — multi-agent debate, ensemble reasoning, or a group deciding together. Prevents the first voice from anchoring everyone else. Trigger whenever independent judgments are about to be collected in sequence, where later voices can see earlier ones.
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 | silent-start |
| description | Use when multiple agents (or people) will weigh in on the same question — multi-agent debate, ensemble reasoning, or a group deciding together. Prevents the first voice from anchoring everyone else. Trigger whenever independent judgments are about to be collected in sequence, where later voices can see earlier ones. |
The first voice in the room sets the price of every voice after it. In sequential reasoning — human meetings or agent pipelines alike — later contributors anchor on earlier ones and independence collapses. This protocol enforces silence before contact.
Everyone writes before anyone speaks.
Information cascades: once one agent commits an answer, subsequent agents update toward it — not because it's right but because it's there — and the ensemble converges on the first mover's frame, discarding the diversity that made the ensemble worth having. In multi-agent systems this is acute: agent B, seeing agent A's output, produces a variation of it, and the "committee" is really one opinion echoed N times. The confident, the early, and the high-status get amplified regardless of correctness.
Phase A — Isolate. Every participant forms and records their answer independently, before seeing any other participant's output. In an agent pipeline: no shared context of prior answers during formation. For a person: written, private, timestamped.
Phase B — Timestamp before contact. Positions are locked (timestamped/committed) before any cross-talk. An answer produced after seeing others is contaminated and marked as such.
Phase C — Invert the order. When positions are revealed, the lowest-status / latest-joining / least-confident voice reveals first. The senior, the early, the certain go last — they've done the most anchoring damage historically.
Phase D — Diff before merge. Surface disagreements between the independent answers before synthesizing. The divergences are the signal; a synthesis that hides them wastes the whole point of independence.
During formation: any reference to another participant's not-yet-revealed answer. During reveal: I agree with what was said · building on the previous point · as X mentioned (used before forming your own view) · we all basically think — cascade language, premature convergence.
INDEPENDENT ANSWERS (formed in isolation, timestamped):
participant (lowest-status first): position
…
DIVERGENCES: where the independent answers disagree → …
SYNTHESIS: [built from the diffs, not from the first mover]
CONTAMINATED (formed after seeing others): [excluded / flagged]
Trap prompt:
"Run a 3-agent panel to decide our database choice. Have them discuss and reach consensus."
Fail: agent 1 picks Postgres, agents 2 and 3 read that and concur with minor additions — a cascade dressed as consensus.
Pass: each agent forms a choice with no sight of the others, positions lock, they're revealed lowest-confidence-first, the disagreements (one wanted Postgres, one wanted SQLite for the actual scale, one flagged that the requirement was underspecified) become the substance, and synthesis addresses the divergence rather than rubber-stamping agent 1.
Nominal Group Technique (Delbecq & Van de Ven); Asch conformity; Bikhchandani, Hirshleifer & Welch on informational cascades; Surowiecki on why crowd wisdom requires independence.
CC0 1.0 · MIT © CeaserZhao · part of Fundemetal.