Imported from benjam3n/reasoningtool (claude-code-plugin/skills/ai/SKILL.md). Install upstream with npx skills add benjam3n/reasoningtool --skill ai. Copyright stays with the author.

Assumption Inversion

Input: $ARGUMENTS

Interpretations

Before executing, identify which interpretation matches the user's input:

Interpretation 1 — Flip assumptions to find blind spots: The user has identified (or wants to identify) assumptions and wants to systematically invert them to reveal hidden risks, overlooked alternatives, and failure modes. Interpretation 2 — Challenge conventional wisdom: The user wants to question widely-held beliefs in a field or organization — "everyone assumes X, but what if X is wrong?" Interpretation 3 — Explore "what if the opposite is true?": The user has a specific belief or premise and wants to think through the consequences of it being exactly wrong, as a creative or strategic exercise.

If ambiguous, ask: "I can help with inverting specific assumptions to find blind spots, challenging conventional wisdom in a domain, or exploring the consequences of a belief being exactly wrong — which fits?" If clear from context, proceed with the matching interpretation.

Purpose

Take extracted assumptions and invert them to discover:

Blind spots in current thinking
Alternative possibilities not considered
Potential failure modes
Novel approaches

Why inversion works:

Assumptions constrain the solution space
Inverting opens new regions of possibility
Many innovations come from questioning "obvious" assumptions

Prerequisites: Assumptions extracted (use /aex first, or provide assumptions directly)

Depth Scaling

Default: 2x. Parse depth from $ARGUMENTS if specified (e.g., "/ai 4x [input]").

Depth	Min Inversions	Min Alternatives per Inversion	Min Stress Tests	Min Depth Levels
1x	3	1	1	2
2x	5	2	2	3
4x	8	3	3	4
8x	12	4	5	5
16x	18	5	8	6

These are floors. Go deeper where insight is dense. Compress where it's not.

The Inversion Process

Step 1: List Assumptions to Invert

ASSUMPTIONS FOR INVERSION:

Source: [where these came from]

1. [Assumption 1]
2. [Assumption 2]
3. [Assumption 3]
...
N. [Assumption N]

TOTAL: [N] assumptions

Step 2: Apply Inversion Techniques

For each assumption, apply multiple inversion types:

Technique	Method	Example
Negation	"What if NOT X?"	"Customers want low prices" -> "What if customers don't want low prices?"
Reversal	"What if the opposite?"	"We sell to customers" -> "What if customers sell to us?"
Elimination	"What if X didn't exist?"	"We need a website" -> "What if we had no website?"
Maximization	"What if X were infinite?"	"Limited budget" -> "What if budget were unlimited?"
Minimization	"What if X were zero?"	"We have a team" -> "What if team size were zero?"
Time shift	"What if X were different timing?"	"Launch next quarter" -> "What if we launched yesterday/in 10 years?"
Actor swap	"What if different actor?"	"We build this" -> "What if competitors/customers built this?"

Step 3: Structured Inversion

For each assumption:

INVERTING: [Assumption]
===================================================

ORIGINAL: [assumption as stated]

INVERSIONS:

1. NEGATION: What if NOT [assumption]?
   -> [inverted statement]
   Implication: [what this would mean]
   Plausibility: [0-100%]

2. REVERSAL: What if the opposite?
   -> [reversed statement]
   Implication: [what this would mean]
   Plausibility: [0-100%]

3. ELIMINATION: What if this didn't exist/matter?
   -> [eliminated version]
   Implication: [what this would mean]
   Plausibility: [0-100%]

4. EXTREME: What if this were 10x or 0.1x?
   -> [extreme version]
   Implication: [what this would mean]
   Plausibility: [0-100%]

MOST INTERESTING INVERSION: [which one]
WHY: [what makes it interesting]

===================================================

Step 4: Filter by Plausibility and Interest

INVERSION TRIAGE:

HIGH PLAUSIBILITY + HIGH INTEREST (explore deeply):
- [Inversion 1]: [brief description]
- [Inversion 2]: [brief description]

HIGH PLAUSIBILITY + LOW INTEREST (note but don't pursue):
- [Inversion 3]: [brief description]

LOW PLAUSIBILITY + HIGH INTEREST (creative exploration):
- [Inversion 4]: [brief description]

LOW PLAUSIBILITY + LOW INTEREST (discard):
- [Inversion 5]: [brief description]

Step 5: Explore Promising Inversions

For each promising inversion, ask:

EXPLORING: [Inverted assumption]
===================================================

IF THIS INVERSION WERE TRUE:

1. What would be different?
   - [Consequence 1]
   - [Consequence 2]
   - [Consequence 3]

2. Who would benefit?
   - [Beneficiary 1]
   - [Beneficiary 2]

3. Who would lose?
   - [Loser 1]
   - [Loser 2]

4. What would we do differently?
   - [Action 1]
   - [Action 2]

5. Is there evidence this is already partially true?
   - [Evidence/counter-evidence]

6. What would make this become true?
   - [Trigger 1]
   - [Trigger 2]

INSIGHT FROM THIS INVERSION:
[Key insight or blind spot revealed]

===================================================

Step 6: Synthesize Blind Spots and Alternatives

===================================================
INVERSION SYNTHESIS: [topic]
===================================================

BLIND SPOTS DISCOVERED:

1. [Blind spot 1]
   Hidden by assumption: [which assumption]
   Revealed by inversion: [which inversion]
   Implication: [what to do about it]

2. [Blind spot 2]
   Hidden by assumption: [which assumption]
   Revealed by inversion: [which inversion]
   Implication: [what to do about it]

===================================================

ALTERNATIVE POSSIBILITIES:

1. [Alternative 1]
   If we assumed: [inverted assumption]
   We could: [new possibility]
   Feasibility: [assessment]

2. [Alternative 2]
   If we assumed: [inverted assumption]
   We could: [new possibility]
   Feasibility: [assessment]

===================================================

FAILURE MODES (inversions that could happen to us):

1. [Failure mode 1]
   If [assumption] becomes false: [consequence]
   Early warning signs: [what to watch]
   Mitigation: [how to prepare]

2. [Failure mode 2]
   If [assumption] becomes false: [consequence]
   Early warning signs: [what to watch]
   Mitigation: [how to prepare]

===================================================

ACTIONABLE INSIGHTS:

1. [Insight 1] -> Action: [what to do]
2. [Insight 2] -> Action: [what to do]
3. [Insight 3] -> Action: [what to do]

===================================================

Quick Inversion (Abbreviated)

For fast blind spot discovery:

QUICK INVERSION: [topic]

Top 3 assumptions:
1. [Assumption 1]
2. [Assumption 2]
3. [Assumption 3]

Quick inversions:
1. What if NOT [A1]? -> [implication]
2. What if opposite of [A2]? -> [implication]
3. What if [A3] didn't exist? -> [implication]

Biggest blind spot: [what we're missing]

Example: "We need to hire more engineers"

Assumptions

More engineers = more output
We need more output
Hiring is the way to get engineers
Engineers are the bottleneck

Inversions

Assumption 1: "More engineers = more output"

NEGATION: More engineers ≠ more output
- Implication: Brooks's Law—adding people to late project makes it later
- Plausibility: 60% (well-documented phenomenon)
REVERSAL: Fewer engineers = more output
- Implication: Small teams move faster, less coordination overhead
- Plausibility: 40% (depends on context)

Assumption 4: "Engineers are the bottleneck"

NEGATION: Engineers are NOT the bottleneck
- Implication: Hiring won't help; need to find real bottleneck
- Plausibility: 50% (often the case)
ELIMINATION: What if there were no bottleneck?
- Implication: System is already optimal; expectations are wrong
- Plausibility: 20%

Synthesis

BLIND SPOTS DISCOVERED:

1. Coordination costs ignored
   Hidden by: "More engineers = more output"
   Revealed by: Negation
   Implication: Calculate coordination cost before hiring

2. Bottleneck assumption untested
   Hidden by: "Engineers are the bottleneck"
   Revealed by: Negation
   Implication: Do bottleneck analysis first

ALTERNATIVE POSSIBILITIES:

1. Improve tooling instead of hiring
   If we assumed: Current engineers are under-leveraged
   We could: Invest in automation, better tools
   Feasibility: High—often better ROI than hiring

2. Reduce scope instead of increasing capacity
   If we assumed: We don't need all planned output
   We could: Prioritize ruthlessly, do less
   Feasibility: Medium—requires stakeholder alignment

FAILURE MODES:

1. If "more = more output" is false
   Consequence: Money wasted, team slower
   Warning signs: Velocity decreasing despite hiring
   Mitigation: Track velocity per engineer

Inversion Patterns

For Business Assumptions

Focus on: Customer assumptions, market assumptions, competitive assumptions

For Technical Assumptions

Focus on: Scalability assumptions, dependency assumptions, architecture assumptions

For Process Assumptions

Focus on: Sequence assumptions, role assumptions, resource assumptions

For Strategy Assumptions

Focus on: Competitive advantage assumptions, market timing assumptions, capability assumptions

Quality Checklist

Before completing:

All assumptions listed
Multiple inversion techniques applied to each
Plausibility and interest rated
Promising inversions explored deeply
Blind spots identified
Alternative possibilities generated
Failure modes mapped
Actionable insights synthesized

Integration

Use with:

/aex -> Extract assumptions first
/cda -> Find domains where inversions are the norm
/ins -> Combine insights from multiple inversions
/araw -> Test inversions with Assume Right / Assume Wrong

ai - Assumption Inversion

About