# Ac Ideation

> Generate and converge on options for a decision: frame before generating, run divergence and convergence separately, and capture each survivor with its assumption and cheapest test. Seven leaves: framing-the-question (decision and decider, walls versus fences), divergence-rules (defer judgment, build on others, silence-then-share, a quota), prompt-techniques (SCAMPER, analogy, inversion, constraints, worst-idea flip, morphology, 10x and 0.1x), nominal-group (silent generation, round-robin, clarification only, ranking), convergence-scoring (criteria before scoring, coarse scales, now/later/drop, kill criteria), idea-capture (one line, a so-that clause, the assumption, cheapest test), session-record (question, constraints, honest count, shortlist, decisions, next test). Use for 'brainstorm this', 'generate options', 'we need ideas', 'how might we', 'get unstuck', 'shortlist these ideas'. Not for attacking an existing plan (ac-interrogation), writing requirements (ac-spec-writing), or deciding alone.

- Skill: `jiayaoqijia/ac-ideation` (Agent Skill, multi-file: 11 files)
- Install (CLI): `npx get-skill add jiayaoqijia/ac-ideation`
- Raw SKILL.md: https://api.skillmd.com/api/skills/jiayaoqijia/ac-ideation/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- License: MIT
- Author: jiayaoqijia (https://skillmd.com/u/jiayaoqijia)
- Updated: 2026-11-16
- Page: https://skillmd.com/skills/jiayaoqijia/ac-ideation

---


# Ideation

## Overview

Two different jobs hide under "brainstorming": generating many options, and cutting them down to a few worth testing. They must not share a meeting, because judgment kills generation. This skill runs both as separate passes: frame the question, diverge, converge, capture, and leave a written record so the session survives its own author.

## When to use

- Someone asks for options: "brainstorm this", "generate options", "we need ideas", "how might we".
- A team or a solo operator is stuck and needs a wider set of alternatives before choosing.
- A long list of ideas needs to be shortlisted honestly, with scores and a next test.

## When NOT to use

Attacking an existing plan is outside; ac-interrogation covers that. Writing requirements is outside; ac-spec-writing covers that. Deciding alone is outside: this skill produces options and a shortlist, never the decision. A routine single-answer question needs one idea, not a session.

## Subskill index

| Leaf | Domain segment | Scope (one line) |
|---|---|---|
| `subskills/framing-the-question.md` | `critical-thinking/framing-the-question` | Turn a topic into a decision brief: the decision, the decider, walls versus fences, and the scope |
| `subskills/divergence-rules.md` | `critical-thinking/divergence-rules` | Rules that raise quantity and variety: defer judgment, build on others, silence-then-share, a quota |
| `subskills/prompt-techniques.md` | `critical-thinking/prompt-techniques` | A technique bank with when-to-use: SCAMPER, analogy, inversion, constraints, worst-idea flip, morphology, 10x |
| `subskills/nominal-group.md` | `critical-thinking/nominal-group` | Structured convergence without anchoring: silent generation, round-robin, clarification only, independent ranking |
| `subskills/convergence-scoring.md` | `critical-thinking/convergence-scoring` | Filtering: criteria before scoring, coarse scales, now/later/two-list, kill criteria, the cheap-test tiebreak |
| `subskills/idea-capture.md` | `critical-thinking/idea-capture` | Normalize each survivor to one line plus a so-that clause, its assumption, and its cheapest test |
| `subskills/session-record.md` | `critical-thinking/session-record` | The record: question, constraints, honest count, clusters, shortlist with scores, decisions, next test |

## Status and honest limits

- Seedling: built from the public canon cited in each leaf; the full pass order has not been stress-tested across many live sessions.
- The skill raises the quantity and quality of options and makes the shortlist defensible. It does not make the choice, and it cannot guarantee a good idea exists in the space searched.
- The quantity rules rest partly on lab studies of face-to-face groups (⚠️, divergence-rules); treat those as directional, not law.
