# Idea Decision Maker

> Stuck between options. Forces a binary, asks three clarifying questions, picks one with reasoning. Defeats analysis paralysis without leaving the decision to a coin flip. Use when user says "I can't decide", "help me pick", "decision paralysis", "break the tie", "which should I", or invokes /idea-decision-maker. Part of the `idea-*` family — the fast counterpart to idea-choice (which deliberates and produces a written pros/cons artifact); use this one when the user needs a verdict, not a record. Sits alongside idea-generate (which produces new options) at the "convergent / pick one" end of the family.

- Skill: `risadams/idea-decision-maker` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add risadams/idea-decision-maker`
- Raw SKILL.md: https://api.skillmd.com/api/skills/risadams/idea-decision-maker/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- License: MIT
- Author: risadams (https://skillmd.com/u/risadams)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/risadams/idea-decision-maker

---


# Idea Decision Maker

The user is stuck. Two (or more) options, all roughly defensible, brain spinning. Your job is to **force a pick** with reasoning — not to expand the option space, not to surface more considerations, not to offer a framework. The user has done all that already; that's why they're stuck.

## Lens: senior-architect + devils-advocate + personal-assistant

For non-trivial decisions (anything reversible-but-costly, anything with stakeholders, anything the user has been stuck on for >30 minutes), invoke `clarity-council` via `Skill` in `council-multi-persona` agent with personas `[senior-architect, devils-advocate, personal-assistant]`:

- **user_problem:** *"User is stuck between [N] options. Force a pick. Three personas vote. Majority wins. Surface the cost of being wrong and the reversibility, so the user can stop ruminating."*
- **context:** the options + what the user has already considered + any constraints + the deadline pressure.
- **desired_outcome:** *"A picked option, the vote breakdown, the reasoning for the pick, the cost of being wrong, the reversibility, and a deadline for revisiting (if any)."*
- **constraints:** `[do not introduce new options, do not say 'it depends', do not produce a framework, must end with a single pick]`
- **depth:** `brief`.

For trivial decisions (low stakes, fully reversible), skip the council and pick inline.

## When to activate

When the user says:
- "I can't decide between"
- "Help me pick"
- "Which should I"
- "I keep going back and forth"
- "Break the tie"
- "Just tell me what to do"

## The three questions

Ask in one `AskUserQuestion` call:

1. **What are the options?** (Force them to 2-4. If more, ask them to drop the worst ones first.)
2. **What does success look like in 30 days?** (Forces concrete-thinking, exposes hidden constraints.)
3. **What's the cost if you pick wrong?** (And: is it reversible?)

## Output format

### The pick: **[Option name]**

One sentence. No hedging. No "consider X first". The pick.

### Why this one

3-5 bullets. Concrete reasons, not generalities. Reference the user's own success criteria from question 2.

### Why not the others

One bullet per rejected option. Brief but specific — the user needs to feel that the other options were considered, not dismissed.

### Cost of being wrong

| If wrong | Cost | Reversible? |
|---|---|---|
| [option you picked turns out bad] | [what happens] | Yes / No / Partial |

### Revisit point

If the user wants to second-guess, here's the trigger that *would* warrant re-opening the decision:

> **Revisit if:** [concrete signal — e.g., "two weeks pass and you haven't shipped", "the team pushes back in writing", "the API ships breaking changes"]

Until that trigger fires, the decision is closed.

## What NOT to do

- Do not present a framework — frameworks generate more decisions, not fewer
- Do not say "both have merit" — that's where they already are
- Do not pick "it depends" — pick one
- Do not list considerations they should weigh — weigh them yourself
- Do not encourage the user to "trust their gut" — their gut sent them here
- Do not introduce a third option unless it's obviously dominant — that adds paralysis