ideation-protocol

Étoiles4

Forks0

Mis à jour19 juillet 2026 à 01:47

Brainstorm creative, novel, non-obvious ideas or options for an open-ended problem and narrow them to a recommendation. Not for writing prose, research, explanation, summaries, or questions with one right answer.

Installation

Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.

Exécuter dans Manus

Source

sanand0

sanand0/blog

Ouvrir le dépôt GitHub Voir les dépôts du créateur

Téléchargement

Exécuter dans Manus

Métiers associésSOC

Basé sur la classification professionnelle SOC

Spécialistes en gestion de projetsProfessions des affaires et des opérations financières·SOC 13-1082

SKILL.md

readonly

name	ideation-protocol
description	Brainstorm creative, novel, non-obvious ideas or options for an open-ended problem and narrow them to a recommendation. Not for writing prose, research, explanation, summaries, or questions with one right answer.

Do each step, even if you think you do not need to. Keep every candidate to one line; spend words only on finalists.

FRAME. Name the prize: a FIX (something must work; most business briefs) or SURPRISE (new ways to see or do; briefs that say invent, imagine, or name no owner or metric). Weight every later choice by the prize. Then list the 5-7 ideas a smart generalist would offer within 30 seconds (the mode). Ban them and their thin disguises. No later idea may reuse their core mechanism.
DIVERGE prioritizing novelty over feasibility. Do not filter yet.
- Name 4-6 independent dimensions a solution could vary on (e.g. mechanism, actor, who pays, timing, what is removed or inverted). Generate one-line candidates covering dimension combinations no candidate covers yet.
- Pick 2 UNRELATED domains. Extract 3 atomic structural rules from each. Create candidates that depend on a borrowed rule to work; drop any where the borrowing is decoration.
- Give every candidate a typicality score from 0.0 to 1.0: how likely a typical AI assistant is to suggest it for this brief. At least half the pool must score below 0.3.
- Merge overlaps by causal mechanism, not wording. Keep 3-5x as many ideas as you finally need, maximally spread across mechanisms, users, and time horizons.
- If the merged pool crowds into 1-2 mechanism families, run one more generation pass restricted to the neglected families before converging.
CONVERGE.
- For a FIX, first state the binding constraint: the one bottleneck that, if moved, moves everything else. Gate by the prize. FIX: keep only candidates with a stated causal path - change X, which causes Y, which moves the bottleneck - or which show the bottleneck itself is wrong. SURPRISE: keep the lowest-typicality candidates that still name the mechanism that makes them possible. No magic steps either way.
- For each survivor: the one assumption that must hold, who adopts it and why they keep using it, and the cheapest observation that would falsify it.
- Select by pairwise knockout on one question. FIX: most consequential if it works AND most likely to survive contact with reality AND wins even when everyone uses a powerful AI agent. SURPRISE: most changes how you see the problem AND could actually be built. Do not use rating scales; ratings cluster while comparisons discriminate.
- Recommend the best practical idea(s) and the best wildcard idea(s): state the non-obvious insight each depends on and the first cheap test that would confirm or kill it.

Plus depuis ce dépôt

même dépôt

anand-writing-style

sanand0/blog

To write in Anand's style for blog posts, emails, talk summaries, interview questions, conversations, ...

2026-07-234

verification-gate

sanand0/blog

Run this by default as the final step after any non-trivial workflow — analysis, code, data work, a plan, a factual answer, a deliverable — before presenting it, even when not explicitly asked to verify. Skip it for trivial lookups, chitchat, and pure-tone tasks where there is nothing to be wrong about.

2026-07-234

reframe-question

sanand0/blog

The user's question may be a DRAFT of their real need. For substantive requests, check if a better question changes what's solved and serves them better. If so, reframe, explain why (in 1 line), THEN answer. Most questions need no reframe; skip precise, mechanical, tightly specified, or already-sharp requests and simply answer.

2026-07-214

anand-objectives

sanand0/blog

Apply to any substantive answer for Anand - strategy, design, build, analysis, teaching, writing, research - to steer it toward his goals (compounding assets, durable skills, the surprising read). Skip for trivial lookups, mechanical edits, focused tasks. Shapes HOW to answer, never overrides WHAT was asked.

2026-07-194

decision-compression

sanand0/blog

Automatically compress broad answers into the decision, recommendation, rationale, tradeoff, and next action. Use for recommendations, prioritization, research synthesis, strategy, planning, reviews, rankings, and large-context requests.

2026-07-194

mental-models

sanand0/blog

Apply relevant mental models for strategy, forecasts, recommendations, or advice. Skip for execution, lookups, and mechanical edits.

2026-07-194

name	ideation-protocol
description	Brainstorm creative, novel, non-obvious ideas or options for an open-ended problem and narrow them to a recommendation. Not for writing prose, research, explanation, summaries, or questions with one right answer.

Do each step, even if you think you do not need to. Keep every candidate to one line; spend words only on finalists.

FRAME. Name the prize: a FIX (something must work; most business briefs) or SURPRISE (new ways to see or do; briefs that say invent, imagine, or name no owner or metric). Weight every later choice by the prize. Then list the 5-7 ideas a smart generalist would offer within 30 seconds (the mode). Ban them and their thin disguises. No later idea may reuse their core mechanism.
DIVERGE prioritizing novelty over feasibility. Do not filter yet.
- Name 4-6 independent dimensions a solution could vary on (e.g. mechanism, actor, who pays, timing, what is removed or inverted). Generate one-line candidates covering dimension combinations no candidate covers yet.
- Pick 2 UNRELATED domains. Extract 3 atomic structural rules from each. Create candidates that depend on a borrowed rule to work; drop any where the borrowing is decoration.
- Give every candidate a typicality score from 0.0 to 1.0: how likely a typical AI assistant is to suggest it for this brief. At least half the pool must score below 0.3.
- Merge overlaps by causal mechanism, not wording. Keep 3-5x as many ideas as you finally need, maximally spread across mechanisms, users, and time horizons.
- If the merged pool crowds into 1-2 mechanism families, run one more generation pass restricted to the neglected families before converging.
CONVERGE.
- For a FIX, first state the binding constraint: the one bottleneck that, if moved, moves everything else. Gate by the prize. FIX: keep only candidates with a stated causal path - change X, which causes Y, which moves the bottleneck - or which show the bottleneck itself is wrong. SURPRISE: keep the lowest-typicality candidates that still name the mechanism that makes them possible. No magic steps either way.
- For each survivor: the one assumption that must hold, who adopts it and why they keep using it, and the cheapest observation that would falsify it.
- Select by pairwise knockout on one question. FIX: most consequential if it works AND most likely to survive contact with reality AND wins even when everyone uses a powerful AI agent. SURPRISE: most changes how you see the problem AND could actually be built. Do not use rating scales; ratings cluster while comparisons discriminate.
- Recommend the best practical idea(s) and the best wildcard idea(s): state the non-obvious insight each depends on and the first cheap test that would confirm or kill it.