Prompt gallery

The best prompt is the one you will actually type. These are real invocations in three styles. The point: a sparse prompt produces the same complete artifact as a polished one - the framework structures the output, not the prompt. You just supply the situation.

The three ways to ask

Style	Who it fits	What it looks like
Casual	you know the move and want it now	one or two sentences, the decision and the worry
Organized	higher stakes, you front-load context	situation, constraints, what you already tried
Advisor-routed	you do not know which framework	describe the mess; let the advisor sequence it

The recurring names below (Mira, Daniel, Priya) are the cast: a founder, an engineer, and a policy analyst, so you can see each style in its natural voice.

Advisor-first: when you do not know the move

Lead here when you are unsure. Describe the situation in plain language to the Framework Advisor and run the plan it returns.

/think-framework-advisor "we keep arguing about whether to rebuild the billing
service or patch it. high stakes, hard to reverse, and honestly I think we're
each defending a position we picked months ago. help me think about it."

Returns a short Thinking Plan: typically frame the real question first, then compare the options, then stress-test the choice - with the one or two frameworks for each step and what to skip.

/think-framework-advisor "stuck on pricing for the new tier. not even sure
pricing is the actual problem."

The advisor will often tell you the framing is the problem before the pricing is - and route you to restate it first.

Premortem

think-premortem - a ranked risk register with tripwires and kill criteria.

Casual (Mira)

/think-premortem "launching a free tier in 6 weeks, nervous it cannibalizes paid
and swamps support"

Organized (Daniel)

/think-premortem "
Decision: cut over the payments service to the new provider in one weekend.
Constraints: 3-engineer team, no rollback once funds route, Q3 freeze starts in 4 weeks.
Already considered: dual-running both providers (too expensive).
Stress-test this plan before we commit."

Advisor-routed

/think-framework-advisor "about to commit to a risky migration, want to know how
it could go wrong before we pull the trigger"

Problem Restatement

think-problem-restatement - several reframings of the problem, with one chosen working frame.

Casual (Mira)

/think-problem-restatement "I can't decide who to hire for the first eng role"

Detailed (Priya)

/think-problem-restatement "
We are framing this as 'how do we increase form completion rates.' Stakeholders:
applicants (mostly mobile, time-poor), caseworkers, the audit team. I suspect the
real problem is upstream of the form. Restate it before we redesign anything."

Comparing options

think-linear-model-aggregation (weighted scoring) and think-pairwise-comparison (head-to-head when you cannot score).

Organized (Daniel), weighted

/think-linear-model-aggregation "
Options: migrate to Postgres, stay on MySQL, move to cloud-hosted.
Criteria: operational cost (high weight), team familiarity (medium), migration
risk (high), query flexibility (low).
Context: 18-month runway, 3-person backend team, no dedicated DBA.
Ruled out: NoSQL (compliance)."

Casual (Mira), head-to-head

/think-pairwise-comparison "three names for the product and I keep flip-flopping.
help me rank them, I don't trust my gut here"

Reasoning through an ethical trade-off

think-ethical-matrix - a grid of affected parties against principles, read for the trade-off pattern.

Detailed (Priya), by hand

/think-ethical-matrix "
Feature: mandatory approval gates on shared documents. Affected: document authors,
approvers, downstream readers, and people who are never consulted but are affected
by what ships. Principles: wellbeing, autonomy, fairness. Surface the trade-offs;
I will run this on paper for the review."

Diagnosing a recurring problem

think-causal-loop-diagrams (feedback loops) and think-process-tracing (rival causes of a single case).

Organized (Daniel)

/think-causal-loop-diagrams "
The approval queue keeps growing no matter how many reviewers we add. Reviewers
say more reviewers means more context-switching means slower reviews. Map the
loops so I can find what actually drives the backlog."

Key takeaway

All three styles, and the advisor route, produce a complete, on-spec artifact. The only difference is how much you front-load and whether you let the advisor choose the sequence. So do not polish - describe the real situation, name the stakes, and run it. To see these prompts produce full artifacts end to end, read the Showcase.