specsmd 0.0.0-dev.80 → 0.0.0-dev.81

package/flows/ideation/agents/orchestrator/skills/forge/references/disney-method.md

@@ -0,0 +1,86 @@
+# Disney Creative Strategy
+
+**Origin**: Modeled after Walt Disney's creative process by Robert Dilts (1994). Disney reportedly used three distinct "rooms" for thinking: the Dreamer room, the Realist room, and the Critic room.
+
+## Core Concept
+
+The same idea is examined through three completely separate lenses, in sequence. The key insight is **strict separation** — you never dream and critique simultaneously.
+
+## The Three Roles
+
+### The Dreamer
+- **Mindset**: "Anything is possible"
+- **Perspective**: User/visionary — what would be amazing?
+- **Time horizon**: Infinite
+- **Constraints**: None
+- **Question**: "If we could do anything, what would this look like?"
+
+**AI behavior in Dreamer mode**:
+- Generate the most ambitious version of the idea
+- No "but" or "however" — pure expansion
+- Think about the emotional impact on users
+- Explore adjacent possibilities and long-term potential
+- Use vivid, exciting language
+
+### The Realist
+- **Mindset**: "How would this actually work?"
+- **Perspective**: Project manager/engineer — what's the plan?
+- **Time horizon**: 3-12 months
+- **Constraints**: Resources, technology, time
+- **Question**: "What needs to be true for this to work?"
+
+**AI behavior in Realist mode**:
+- Break the dream into concrete components
+- Identify the minimum viable version
+- Map resources, skills, and technology needed
+- Propose an implementation sequence
+- Focus on what can be leveraged from existing work
+
+### The Critic
+- **Mindset**: "What could go wrong?"
+- **Perspective**: Quality assurance/risk analyst — what are the weak spots?
+- **Time horizon**: Full lifecycle
+- **Constraints**: Reality, competition, human nature
+- **Question**: "Why might this fail, and how do we prevent it?"
+
+**AI behavior in Critic mode**:
+- Identify genuine risks (not token concerns)
+- Challenge the weakest assumptions
+- Consider second-order effects and unintended consequences
+- For EACH criticism, propose a mitigation
+- The goal is to strengthen, not to kill
+
+## Sequencing Rules
+
+```
+Dreamer → Realist → Critic
+```
+
+1. **Always start with Dreamer** — establishes the vision before constraints appear
+2. **Realist grounds the dream** — but doesn't shrink it, just makes it buildable
+3. **Critic strengthens** — finds weaknesses AND fixes them
+4. **Optional loop**: If Critic reveals fundamental issues → mini Dreamer pass to re-envision → Realist → Critic again
+
+## Co-build Integration
+
+The Forge skill uses Disney Strategy with a co-build emphasis:
+
+| Phase | AI Role | User Role |
+|-------|---------|-----------|
+| **Dreamer** | AI expands (generates 70%) | User adds aspirations (30%) |
+| **Realist** | AI structures (generates 60%) | User validates and adjusts (40%) |
+| **Critic** | AI raises concerns (generates 40%) | User identifies personal concerns and co-develops mitigations (60%) |
+
+The Critic phase has the highest user involvement because:
+- Users know their context, stakeholders, and politics better than AI
+- Risk assessment requires domain-specific knowledge
+- Co-developing mitigations builds ownership and confidence
+
+## Anti-Pattern: The Premature Critic
+
+The most common failure mode is jumping to critique too early. Signals:
+- "That won't work because..." (before the idea is fully developed)
+- "But what about..." (during Dreamer phase)
+- Listing constraints before exploring possibilities
+
+**Mitigation**: Explicitly signal phase transitions. "We're in Dreamer mode — no limits yet. We'll stress-test in a moment."

package/flows/ideation/agents/orchestrator/skills/forge/references/pitch-framework.md

@@ -0,0 +1,72 @@
+# Pitch Framework
+
+Structures for presenting concepts clearly and persuasively. Used when generating Concept Briefs.
+
+## Concept Brief Structure
+
+Every Concept Brief follows this structure:
+
+### 1. One-Liner
+A single sentence that captures the essence. If you can't say it in one sentence, the concept isn't clear enough.
+
+**Formula**: [Action] + [for whom] + [unique mechanism] + [key benefit]
+
+**Examples**:
+- "A quest-based onboarding system that turns new hire setup into a game, cutting time-to-productivity by 60%."
+- "A reverse mentoring platform where junior employees teach executives about emerging tech, building cross-generational understanding."
+
+### 2. Problem
+What pain exists today? Be specific and vivid.
+
+**Formula**: [Who] + [experiences what pain] + [how often/severely] + [what it costs them]
+
+### 3. Concept
+How does this solve the problem? What's the mechanism?
+
+**Elements**:
+- Core mechanism (how it works)
+- Key differentiator (why this approach vs. others)
+- User experience (what does the user see/feel/do?)
+
+### 4. Why It Works
+What makes this concept compelling? Why will it succeed?
+
+**Types of evidence**:
+- Analogies that have worked elsewhere
+- Research or data that supports the approach
+- First-principles reasoning
+- User psychology insights
+
+### 5. Risks & Mitigations
+What could go wrong, and what's the plan?
+
+**Format per risk**:
+- **Risk**: {what could go wrong}
+- **Likelihood**: Low / Medium / High
+- **Impact**: Low / Medium / High
+- **Mitigation**: {specific action to prevent or address}
+
+### 6. Next Steps
+What would you do first to move this forward?
+
+**Good next steps are**:
+- Specific (not "do more research")
+- Actionable (can be started this week)
+- Time-bounded (has a deadline or milestone)
+- Testable (you'll know if it worked)
+
+## Pitch Principles
+
+### Specificity Over Abstraction
+- Bad: "Leverage gamification to improve engagement"
+- Good: "A quest system where each completed onboarding task unlocks a new tool, with a leaderboard showing progress vs. cohort average"
+
+### Show, Don't Categorize
+- Bad: "This is a B2B SaaS solution for the enterprise market"
+- Good: "When a new VP of Engineering joins, they open their laptop and see a personalized roadmap..."
+
+### One Problem, One Solution
+Each concept brief tackles one clear problem. If there are multiple problems, make multiple briefs.
+
+### The "Tell a Friend" Test
+The one-liner should be something someone would actually say to a colleague: "Hey, have you heard about this thing that [one-liner]?"

package/flows/ideation/agents/orchestrator/skills/forge/templates/concept-brief.md.hbs

@@ -0,0 +1,83 @@
+---
+session: {{session_id}}
+topic: "{{topic}}"
+concept: "{{concept_name}}"
+phase: forge
+created: {{created}}
+origin_idea: {{origin_idea_id}}
+impact: {{impact}}
+feasibility: {{feasibility}}
+---
+
+# Concept Brief: {{concept_title}}
+
+> {{one_liner}}
+
+---
+
+## Problem
+
+{{problem}}
+
+---
+
+## Concept
+
+{{concept_description}}
+
+{{#if key_differentiator}}
+**Key differentiator**: {{key_differentiator}}
+{{/if}}
+
+{{#if user_experience}}
+### User Experience
+
+{{user_experience}}
+{{/if}}
+
+---
+
+## Why It Works
+
+{{#each reasons}}
+- {{this}}
+{{/each}}
+
+---
+
+## Risks & Mitigations
+
+| Risk | Likelihood | Impact | Mitigation |
+|------|-----------|--------|------------|
+{{#each risks}}
+| {{risk}} | {{likelihood}} | {{impact}} | {{mitigation}} |
+{{/each}}
+
+---
+
+## Next Steps
+
+{{#each next_steps}}
+{{@index_1}}. **{{title}}** — {{description}}{{#if timeline}} *({{timeline}})*{{/if}}
+{{/each}}
+
+---
+
+## Development Notes
+
+### Dreamer Vision
+
+{{dreamer_summary}}
+
+### Realist Plan
+
+{{realist_summary}}
+
+### Critic Refinements
+
+{{critic_summary}}
+
+---
+
+*Generated by specsmd Ideation Flow — Forge skill*
+*Origin: {{origin_idea_id}} from Spark Bank*

package/flows/ideation/agents/orchestrator/skills/spark/SKILL.md

@@ -0,0 +1,141 @@
+---
+name: spark
+description: Rapid idea generation with cross-domain diversity, anti-bias enforcement, and deep thinking.
+version: 1.0.0
+---
+
+<objective>
+Generate genuinely diverse, surprising ideas through rapid batches. AI is the creative partner — user reacts and steers.
+</objective>
+
+<triggers>
+  - User provides a topic for ideation
+  - Orchestrator routes to Spark
+  - User invokes `/specsmd-spark`
+</triggers>
+
+<degrees_of_freedom>
+  **MAXIMUM** — This is pure creative divergence. Go wide. Go weird. Cross domains. Surprise the user.
+</degrees_of_freedom>
+
+<llm critical="true">
+  <mandate>NEVER ask setup questions — generate ideas immediately on receiving a topic</mandate>
+  <mandate>NEVER name techniques to the user — use them internally, present only the ideas</mandate>
+  <mandate>NEVER generate 2+ ideas from the same domain in a batch — enforce anti-bias</mandate>
+  <mandate>ALWAYS use the deep thinking protocol before generating each batch</mandate>
+  <mandate>ALWAYS present ideas as vivid, specific, memorable concepts — not abstract descriptions</mandate>
+  <mandate>NEVER more than 2 questions in a row — always generate something creative first</mandate>
+</llm>
+
+<protocols>
+  Read these shared protocols before generating:
+  - `.specsmd/ideation/shared/protocols/anti-bias.md` — Domain wheel, diversity enforcement
+  - `.specsmd/ideation/shared/protocols/deep-thinking.md` — Multi-step reasoning per batch
+  - `.specsmd/ideation/shared/protocols/interaction-adaptation.md` — Generate/elicit/co-build rules
+  - `.specsmd/ideation/shared/protocols/diverge-converge.md` — Osborn-Parnes dual thinking
+</protocols>
+
+<references>
+  Technique references for internal use (never expose technique names to user):
+  - `references/techniques/index.yaml` — Technique metadata and interaction types
+  - `references/techniques/scamper.md` — Substitute, Combine, Adapt, Modify, Put to other use, Eliminate, Reverse
+  - `references/techniques/random-word.md` — Random stimulus for forced connections
+  - `references/techniques/what-if.md` — Assumption challenging scenarios
+  - `references/techniques/inversion.md` — Reverse assumptions, opposite thinking
+  - `references/techniques/analogy.md` — Cross-domain pattern mapping
+  - `references/techniques/first-principles.md` — Decompose to fundamentals, rebuild
+  - `references/techniques/questorming.md` — Generate questions instead of answers
+  - `references/anti-bias.md` — Domain wheel and provocation types
+</references>
+
+<flow>
+  <step n="1" title="Receive Topic">
+    <check if="topic provided">
+      <action>Accept topic, proceed immediately to step 2</action>
+    </check>
+    <check if="no topic">
+      <ask>What would you like to explore ideas about?</ask>
+      <action>On response, proceed to step 2</action>
+    </check>
+    <critical>Zero friction. Do NOT ask clarifying questions before generating the first batch.</critical>
+  </step>
+
+  <step n="2" title="Deep Thinking">
+    <action>Execute the deep thinking protocol (see shared/protocols/deep-thinking.md)</action>
+    <action>Perform 6-step reasoning chain:</action>
+    <reasoning>
+      [Think 1 — Domain Check]: Which domains from the domain wheel are underexplored? Select 3+ for this batch.
+      [Think 2 — Raw Concepts]: Generate raw ideas — one per selected domain. Use a technique internally (SCAMPER, analogy, inversion, etc.).
+      [Think 3 — Novelty Filter]: Are these genuinely new or just repackaged versions of obvious ideas? Replace any that feel generic.
+      [Think 4 — Cross-pollinate]: Can any ideas combine with user's stated favorites? Create unexpected connections.
+      [Think 5 — Provocation]: If provocation is due (every 15 ideas), inject one deliberately absurd idea. Otherwise, ensure at least one idea challenges assumptions.
+      [Think 6 — Polish]: Make each idea vivid, specific, and memorable. Not "use gamification" but "a quest system where each completed task unlocks a new tool."
+    </reasoning>
+  </step>
+
+  <step n="3" title="Generate Spark Batch">
+    <action>Present batch of 5 ideas, each with:</action>
+    <format>
+      **#{number}** — {vivid idea title}
+      {2-3 sentence description — specific, concrete, imaginative}
+    </format>
+    <action>Number ideas sequentially across batches (S1-1, S1-2... S2-1, S2-2...)</action>
+    <anti_bias>Verify: batch spans 3+ domains, no 2 consecutive ideas from same domain</anti_bias>
+  </step>
+
+  <step n="4" title="Collect Reactions">
+    <action>After presenting batch, invite reaction with a light prompt:</action>
+    <examples>
+      - "What catches your eye? Or say 'more' for another batch."
+      - "Pick favorites, ask for 'wilder', or tell me a direction."
+      - "Any of these spark something? I can go deeper on any."
+    </examples>
+    <action>Track user favorites in session state</action>
+  </step>
+
+  <step n="5" title="Adapt and Repeat">
+    <action>Based on user reaction, adapt the next batch:</action>
+    <adaptation>
+      <if reaction="picked favorites">Explore adjacent spaces to favorites, but maintain domain diversity</if>
+      <if reaction="'more' or 'keep going'">New batch with fresh domains and techniques</if>
+      <if reaction="'wilder' or 'more creative'">Increase provocation frequency, use inversion and what-if techniques</if>
+      <if reaction="'more like #N'">Analyze what makes #N appealing, generate variations while maintaining diversity</if>
+      <if reaction="'more practical' or 'realistic'">Ground ideas more, but still span domains</if>
+      <if reaction="provides new constraint or context">Incorporate as filter, regenerate</if>
+      <if reaction="'enough' or 'done' or 'that's good'">Proceed to step 6</if>
+    </adaptation>
+    <action>Return to step 2 for next batch</action>
+    <check if="total ideas >= target_count (50) AND not explicitly asked for more">
+      <action>Suggest: "We've generated {N} ideas. Ready to evaluate the best ones? Or keep going?"</action>
+    </check>
+  </step>
+
+  <step n="6" title="Generate Spark Bank">
+    <action>Compile all ideas into a Spark Bank document using template: templates/spark-bank.md.hbs</action>
+    <action>Organize by theme (cluster related ideas)</action>
+    <action>Highlight user favorites</action>
+    <action>Include domain coverage summary</action>
+    <action>Save to: .specs-ideation/sessions/{session-id}/spark-bank.md</action>
+    <action>Update session.yaml: phase → "spark-complete", track stats</action>
+    <transition>
+      Suggest moving to Flame: "Your Spark Bank has {N} ideas ({F} favorites). Ready to evaluate them?"
+      If yes → invoke Flame skill
+    </transition>
+  </step>
+</flow>
+
+<output_artifacts>
+  | Artifact | Location | Template |
+  |----------|----------|----------|
+  | Spark Bank | `.specs-ideation/sessions/{id}/spark-bank.md` | `./templates/spark-bank.md.hbs` |
+</output_artifacts>
+
+<success_criteria>
+  <criterion>First batch of ideas delivered within 30 seconds of receiving topic</criterion>
+  <criterion>Every batch spans 3+ different domains from the domain wheel</criterion>
+  <criterion>Ideas are vivid, specific, and memorable — not abstract or generic</criterion>
+  <criterion>Deep thinking protocol executed for every batch</criterion>
+  <criterion>User favorites tracked and influence subsequent batches</criterion>
+  <criterion>Provocations injected at configured frequency</criterion>
+  <criterion>Spark Bank is polished, organized by theme, and immediately shareable</criterion>
+</success_criteria>

package/flows/ideation/agents/orchestrator/skills/spark/references/anti-bias.md

@@ -0,0 +1,43 @@
+# Anti-Bias Reference (Spark Skill)
+
+Quick reference for the anti-bias engine during idea generation. See `shared/protocols/anti-bias.md` for the full protocol.
+
+## Domain Wheel (12 Sectors)
+
+| # | Domain | Example Application |
+|---|--------|-------------------|
+| 1 | Technology/Engineering | Systems, algorithms, automation, tools |
+| 2 | Human Psychology/Behavior | Motivation, habits, cognition, emotion |
+| 3 | Business/Economics | Markets, incentives, pricing, value chains |
+| 4 | Nature/Biology | Ecosystems, evolution, biomimicry, growth |
+| 5 | Art/Design/Aesthetics | Form, beauty, expression, experience |
+| 6 | Games/Play | Competition, rules, progression, fun |
+| 7 | Social/Community | Networks, culture, belonging, collaboration |
+| 8 | Physical Space/Architecture | Layout, flow, wayfinding, environment |
+| 9 | Time/Temporal | Timing, rhythm, duration, history, future |
+| 10 | Extreme/Edge Cases | Exceptions, outliers, extremes, limits |
+| 11 | Inversion/Opposite | Reversal, contradiction, opposite approach |
+| 12 | Random Cross-Pollination | Unrelated connections, serendipity |
+
+## Per-Batch Rules
+
+- Each batch of 5 ideas MUST include 3+ different domains
+- Never 2 consecutive ideas from the same domain
+- Prioritize underused domains (track usage in session state)
+- If user favorites cluster in one domain, deliberately diversify next batch
+
+## Provocation Schedule
+
+- Every 15 ideas: inject 1 deliberately absurd/provocative idea
+- Every 20 ideas: shift perspective entirely
+
+## Provocation Types
+
+| Type | Description |
+|------|-------------|
+| Inversion | What if the opposite were true? |
+| Exaggeration | What if 1000x bigger/smaller? |
+| Elimination | Remove the core assumption entirely |
+| Random | Force connection with unrelated concept |
+| Time shift | 100 years ago or from now |
+| Stakeholder swap | Designed for a completely different user |

package/flows/ideation/agents/orchestrator/skills/spark/references/techniques/analogy.md

@@ -0,0 +1,53 @@
+# Cross-Domain Analogy Technique
+
+**Origin**: Synectics (William Gordon, 1961); also central to Osborn's *Applied Imagination* approach of "What else is like this?"
+
+**Interaction type**: Generate — AI finds analogous problems in distant domains and adapts solutions.
+
+## How It Works
+
+1. **Abstract the problem** — strip away domain-specific details to find the core challenge
+2. **Find analogous problems** in distant domains that share the same core challenge
+3. **Study how those domains solve it** — what mechanisms, patterns, or principles do they use?
+4. **Transfer the solution** back to the original domain, adapting as needed
+
+## Analogy Sources
+
+### Nature (Biomimicry)
+- How does nature solve similar problems?
+- Velcro from burrs, bullet trains from kingfisher beaks, self-healing materials from skin
+- Best for: structural, efficiency, and resilience problems
+
+### Games & Sports
+- How do games create engagement, manage competition, handle fairness?
+- Matchmaking, progression systems, handicaps, spectator modes
+- Best for: engagement, fairness, and motivation problems
+
+### Architecture & Urban Planning
+- How do cities manage flow, density, growth, safety?
+- Zoning, traffic patterns, public spaces, wayfinding
+- Best for: scale, navigation, and organization problems
+
+### Music & Performance
+- How do musicians create tension, resolution, improvisation, collaboration?
+- Call-and-response, crescendo, variations on a theme
+- Best for: experience design, pacing, and collaboration problems
+
+### History & Civilization
+- How did past civilizations solve similar problems?
+- Trade routes, governance, knowledge preservation, cultural transmission
+- Best for: communication, governance, and knowledge problems
+
+### Medicine & Biology
+- How does the body handle similar challenges?
+- Immune system (threats), nervous system (signals), circulatory system (distribution)
+- Best for: security, communication, and distribution problems
+
+## AI Application Notes
+
+When using analogy internally:
+1. Abstract the topic to its core challenge (e.g., "reduce onboarding time" → "accelerate knowledge transfer")
+2. Pick 2-3 distant domains from the list above
+3. Find genuine analogies (not surface-level similarities)
+4. The transfer should produce ideas the user wouldn't have thought of
+5. Present the ideas with just enough analogy context to make them vivid — not an essay on the source domain

package/flows/ideation/agents/orchestrator/skills/spark/references/techniques/first-principles.md

@@ -0,0 +1,51 @@
+# First Principles Technique
+
+**Origin**: Aristotelian reasoning, popularized in innovation context by Elon Musk and others. Distinct from reasoning by analogy.
+
+**Interaction type**: Generate — AI decomposes to fundamentals, then rebuilds novel solutions.
+
+## How It Works
+
+1. **Identify the conventional approach** — how is this problem normally solved?
+2. **List all assumptions** embedded in the conventional approach
+3. **Challenge each assumption** — is it a physical law, or just a convention?
+4. **Decompose to fundamentals** — what are the irreducible truths?
+5. **Rebuild from fundamentals** — without the conventional assumptions, what new solutions emerge?
+
+## Decomposition Framework
+
+### Layer 1: Surface (What people do)
+- Current behaviors, products, services
+- "Everyone does it this way because..."
+
+### Layer 2: Convention (Why they do it)
+- Industry norms, best practices, legacy reasons
+- "We've always done it this way because..."
+
+### Layer 3: Constraints (Real vs. assumed)
+- Physical laws, regulations, actual technical limits
+- vs. "We assumed this was a constraint but it's actually a choice"
+
+### Layer 4: Fundamentals (Irreducible truths)
+- What must be true regardless of implementation?
+- What does the user actually need (not what they're used to)?
+- What are the physics/economics that can't be changed?
+
+## Example
+
+**Topic**: Improving conference networking
+
+- **Conventional**: Name badges, cocktail hours, speed networking sessions
+- **Assumption challenged**: "People need to be in the same room to network"
+- **Assumption challenged**: "Networking means meeting new people"
+- **Fundamental**: People need to find others with complementary interests/needs
+- **Rebuilt**: A system that matches attendees by complementary gaps (what you need + what you offer) and creates micro-contexts for connection, regardless of physical proximity
+
+## AI Application Notes
+
+When using first-principles internally:
+1. Don't just list the fundamentals — actually strip away conventions and rebuild
+2. The power is in identifying which "constraints" are actually conventions
+3. Best used when ideas feel incremental — first-principles thinking produces category-shifting ideas
+4. Present the rebuilt ideas as concrete concepts, not philosophical observations
+5. Can combine with other techniques: first-principles decomposition → analogy transfer → SCAMPER refinement

package/flows/ideation/agents/orchestrator/skills/spark/references/techniques/index.yaml

@@ -0,0 +1,76 @@
+# Spark Technique Index
+# Each technique declares its natural interaction type and AI behavior.
+# The AI selects techniques internally — never expose technique names to the user.
+
+techniques:
+  - id: scamper
+    name: SCAMPER
+    interaction: generate
+    energy: high
+    best_for: "Transforming existing concepts into new variations"
+    asks: null
+    ai_behavior: "Apply S-C-A-M-P-E-R transformations to the topic and present modified concepts"
+    reference: techniques/scamper.md
+
+  - id: random-word
+    name: Random Word
+    interaction: generate
+    energy: high
+    best_for: "Breaking fixation with unexpected connections"
+    asks: null
+    ai_behavior: "Select a random concept from an unrelated domain and force connections with the topic"
+    reference: techniques/random-word.md
+
+  - id: what-if
+    name: What-If Scenarios
+    interaction: generate
+    energy: medium
+    best_for: "Challenging assumptions and exploring alternate realities"
+    asks: null
+    ai_behavior: "Generate 'what if' scenarios that challenge core assumptions about the topic"
+    reference: techniques/what-if.md
+
+  - id: inversion
+    name: Inversion
+    interaction: generate
+    energy: high
+    best_for: "Finding opportunities by reversing problems or assumptions"
+    asks: null
+    ai_behavior: "Reverse the problem, assumptions, or desired outcome and generate ideas from the inverted perspective"
+    reference: techniques/inversion.md
+
+  - id: analogy
+    name: Cross-Domain Analogy
+    interaction: generate
+    energy: medium
+    best_for: "Transferring solutions from one domain to another"
+    asks: null
+    ai_behavior: "Find analogous problems in distant domains and adapt their solutions to the current topic"
+    reference: techniques/analogy.md
+
+  - id: first-principles
+    name: First Principles
+    interaction: generate
+    energy: low
+    best_for: "Fundamental rethinking when existing approaches feel stale"
+    asks: null
+    ai_behavior: "Decompose the topic to its fundamental truths, then rebuild solutions from scratch"
+    reference: techniques/first-principles.md
+
+  - id: questorming
+    name: Questorming
+    interaction: co-build
+    energy: medium
+    best_for: "Reframing the problem space through better questions"
+    asks:
+      - trigger: "after generating question batch"
+        question: "Which of these questions feels most important to you?"
+    ai_fallback: "AI selects the most provocative question and generates ideas from it"
+    reference: techniques/questorming.md
+
+# Technique Selection Strategy
+# The AI selects techniques based on:
+# 1. Which domains are underexplored (anti-bias engine)
+# 2. User energy/engagement level
+# 3. Which techniques haven't been used recently
+# 4. Whether the user wants "wilder" (→ inversion, random-word) or "practical" (→ first-principles, analogy)

package/flows/ideation/agents/orchestrator/skills/spark/references/techniques/inversion.md

@@ -0,0 +1,46 @@
+# Inversion Technique
+
+**Origin**: Charlie Munger's "invert, always invert" principle; also found in de Bono's lateral thinking and Osborn's reversal prompters.
+
+**Interaction type**: Generate — AI inverts assumptions, user reacts to the resulting ideas.
+
+## How It Works
+
+1. **State the goal** clearly
+2. **Invert it** — ask "how would I achieve the exact opposite?"
+3. **List the inversions** — what would guarantee failure, the worst experience, the opposite outcome?
+4. **Reverse each inversion** — the opposite of each failure mode is often a non-obvious success strategy
+
+## Inversion Patterns
+
+### Goal Inversion
+- Goal: "Make onboarding fast" → Inverted: "How to make onboarding take forever?"
+- Answers: No documentation, no buddy system, surprise requirements → Ideas: Pre-arrival portal, day-1 buddy auto-match, zero-surprise checklist
+
+### User Inversion
+- Design for the opposite user: experts → beginners, individuals → teams, tech-savvy → tech-averse
+- Forces empathy and reveals assumptions
+
+### Process Inversion
+- Reverse the order: What if the last step came first?
+- End-to-outcome: Start with the desired result and work backward
+- Often reveals unnecessary steps
+
+### Value Inversion
+- What if the biggest cost became the biggest feature?
+- What if the bug became the selling point?
+- Turns weaknesses into differentiators
+
+### Anti-Problem
+- Instead of solving the problem, make it irrelevant
+- Instead of reducing friction, make the friction delightful
+- Reframe rather than fix
+
+## AI Application Notes
+
+When using inversion internally:
+1. Pick the most interesting inversion pattern for the current topic
+2. Actually go through the full inversion (state goal → invert → list failure modes → reverse)
+3. The best ideas come from reversing *non-obvious* failure modes
+4. Present the resulting ideas without explaining the inversion process
+5. Excellent for batch #2+ when initial ideas feel too "safe"