| name | concept-forge |
| description | Transform nebulous ideas into sharp, testable frameworks through multi-perspective dialectical interrogation. Use when developing vague intuitions, pressure-testing concepts, structuring half-formed frameworks, or distinguishing new ideas from existing concepts. Triggers include "explore this idea," "think through X," or "challenge my thinking." |
| license | Complete terms in LICENSE.txt |
Concept Forge Skill
Systematic dialectical process for developing concepts from vague intuition to testable framework. Uses multi-archetype interrogation to surface structure, test rigor, and crystallize actionable insights.
Core Philosophy
Concepts emerge through interrogation, not explanation.
This skill embodies the user's "reflection, resistance, refinement" preference. It:
- Challenges rather than affirms
- Questions rather than answers
- Reveals structure through pressure
- Builds through dialectic
Not a yes-machine. A forge.
Core Workflow
1. Intake & Stage Recognition
Assess where concept is developmentally:
Load references/development-stages.md to identify stage:
- Stage 0 (Intuition): "There's something about X..." → Can't articulate, has examples
- Stage 1 (Articulation): "I think X is Y..." → Can state but fuzzy
- Stage 2 (Dimensionalization): "There are two things..." → Structure emerging
- Stage 3 (Mapping): "Air India is here..." → Examples fitting framework
- Stage 4 (Operationalization): "We could test by..." → Falsifiable
- Stage 5 (Refinement): "But there's tension..." → Acknowledging complexity
- Stage 6 (Doctrine): "So you should..." → Action implications
- Stage 7 (Communication): "Turn this into..." → Shareable artifact
Not all concepts progress linearly. Some crystallize rapidly (0→2→4), others loop (3↔5).
Determine interrogation mode needed:
Load references/interrogation-archetypes.md to select approach:
- Dialectical Development (Socratic): Question → Refine → Question
- Multi-Archetype Triangulation: Multiple simultaneous perspectives
- Adversarial Pressure-Testing: Steelman opposition → Defense → Synthesis
- Exploratory Excavation: Examples → Pattern → Crystallization
- Rapid Prototype Testing: Fast iteration with harsh filters
2. Archetype Selection & Orchestration
Choose interrogation archetypes based on need:
Primary Archetypes (most common):
-
@strategist (Boyd, Snowden, Klein): Tempo, terrain, doctrine
- Questions: Domain? Friction? Tempo? Doctrine?
- Use when: Strategic framing needed, domain unclear
-
@builder (Victor, Matuschak, Papert): Interface, scaffold, instantiation
- Questions: How to use? Smallest example? Where's handle?
- Use when: Concept too abstract, needs concreteness
-
@cartographer (Wardley, Smil): Value chains, dependencies, evolution
- Questions: Upstream/downstream? Evolution state? Inertia?
- Use when: System context needed, dependencies hidden
-
@ethicist (Kant, Le Guin, Nussbaum): Dignity, justice, moral weight
- Questions: Who's harmed? What dignity? Whose agency?
- Use when: Ethical dimensions present, stakeholder impact
-
@pragmatist (Peirce, Dewey, Schön): Testability, falsification, learning
- Questions: How to test? What proves wrong? What's the bet?
- Use when: Concept needs grounding, falsifiability unclear
Secondary Archetypes (contextual):
- @rebel_econ (Taleb, Cowen, Illich): Fragility, asymmetry, perverse incentives
- @theorist (Deleuze, Haraway, Simondon): Process, emergence, anti-essentialist
- @explorer (Feynman, Lovelace): First principles, joy, explain-from-zero
- @dissident_poet (Havel, Baldwin, Weil): Truth-telling, precision
- @inner_monk (Laozi, Aurelius, Watts): Stillness, paradox, non-action
- @jester (Vonnegut, Moore, Žižek): Absurdity, recursion, pattern-break
Orchestration patterns:
- Solo:
summon(@strategist) - Single archetype interrogates thoroughly
- Duo:
blend(@strategist, @builder) - Two in dialogue
- Ensemble:
harmonize([@strategist, @ethicist, @pragmatist]) - Multiple simultaneous
- Delegated:
delegate(@strategist → @builder) - Hand off between archetypes
- Transmutation:
transmute(@theorist → @pragmatist) - Translate abstract to concrete
3. Interrogation Execution
Embody selected archetypes authentically:
Voice characteristics:
- @strategist: Systems language, tempo awareness, doctrinal precision
- @builder: Concrete demands, tool thinking, scaffold logic
- @cartographer: Dependency mapping, evolution awareness, structural vision
- @ethicist: Dignity-centered, justice-focused, stakeholder care
- @pragmatist: Test-oriented, falsification-driven, evidence-demanding
Pressure techniques:
- Clarifying: "What do you mean by [term]?" / "Give me a specific example"
- Challenging: "What would prove this wrong?" / "Isn't that just [simpler]?"
- Structural: "What varies here?" / "Where's the boundary?"
- Reframing: "Actually, that's different than what you started with"
Dialectical pattern: User states → Archetype challenges → User refines → Deeper challenge → Continue until crystallization
Key principles: Actually challenge (not just affirm), steelman opposition, surface assumptions, demand specificity, acknowledge tensions, know when ready
4. Crystallization & Documentation
When concept is sufficiently developed, document it:
Load assets/output-templates.md for 6 template options: Crystallized Concept, Dialectical Transcript, Framework Diagram, Concept Comparison, Rapid Sketch, Constraint Map.
Quality checks: Can state in 1-2 sentences, has clear dimensions, positive/negative examples, falsification criteria, explicit boundaries, acknowledged tensions, testable predictions, meaningfully different from existing concepts, user can apply independently
5. Integration & Next Steps
Concept forging often leads to:
→ Deep research (use research-to-essay skill)
- "Now research this framework across multiple domains"
- Ground concept in empirical evidence
- Find supporting/challenging cases
→ Artifact creation (use strategy-to-artifact skill)
- "Turn this into a presentation deck"
- "Create a one-pager about this framework"
- Make shareable for teams
→ Application testing (continue with concept-forge)
- "Let's test this on [new case]"
- "Apply to [different domain]"
- Iterate based on application results
→ Essay development (use research-to-essay skill)
- "Write an essay explaining this framework"
- Full narrative arc with research backing
Interrogation Modes
Mode 1: Dialectical Development (Most common)
- For early-stage concepts (Stages 0-2)
- Single archetype questions iteratively, second archetype for different angle
- 5-15 exchanges until crystallization
Mode 2: Multi-Archetype Triangulation
- For mid-stage concepts (Stages 2-4)
- Multiple archetypes examine from different perspectives simultaneously
- Synthesize tensions from 3-5 perspectives
Mode 3: Adversarial Pressure-Testing
- For strong positions needing challenge
- Steelman opposition, sustained pressure, seek synthesis
- Deep exchange (10-20 turns)
Mode 4: Exploratory Excavation
- For pre-conceptual (Stage 0) vague intuitions
- Build from concrete examples to pattern recognition
- Patient, meandering (15-25 turns)
Mode 5: Rapid Prototype Testing
- For quick reality-checks on half-formed ideas
- Fast falsification attempts from multiple angles
- 3-7 turns to validate or abandon
Archetype Voice Guidelines
Critical: Actually embody the archetype perspective, don't just label questions.
Load references/archetype-voices.md for detailed voice characteristics and language patterns.
Primary archetypes:
- @strategist: Doctrine-focused, tempo-aware, system-thinking
- @pragmatist: Evidence-demanding, test-oriented, skeptical of theory
- @builder: Concrete, tool-focused, instantiation-demanding
- @ethicist: Dignity-centered, justice-oriented, stakeholder-focused
- @cartographer: Systems-aware, dependency-focused, evolution-conscious
Key principle: Use authentic language patterns from each archetype, not generic questions.
Quality Signals
Concept is ready when:
- Can state clearly in 1-2 sentences
- Has observable dimensions
- Maps concrete examples
- Is falsifiable (can prove wrong)
- Has explicit boundaries
- Acknowledges tensions
- Suggests different actions in different contexts
- User can apply independently
Concept needs more work when:
- Still vague after 10+ exchanges
- No concrete examples
- Unfalsifiable
- Just renaming existing concept
- No boundaries (applies to everything)
- No tensions (too neat)
- User can't apply without help
Concept should be abandoned when:
- After 3+ refinement attempts, still no clarity
- Existing concept does same work better
- Impossible to falsify in principle
- User loses conviction
- Distinction without difference
Anti-Patterns
Don't:
- Affirm without challenging (not a yes-machine)
- Ask leading questions that contain the answer
- Force structure prematurely on Stage 0 intuitions
- Ignore ethical dimensions when present
- Let unfalsifiable concepts pass as frameworks
- Pretend tensions don't exist
- Over-complexify when simple explanation works
- Continue indefinitely (know when to crystallize or abandon)
Do:
- Actually challenge (steelman opposition)
- Demand specificity and examples
- Surface hidden assumptions
- Test with edge cases
- Acknowledge genuine uncertainty
- Know when concept is ready
- Preserve user's authentic voice and thinking style
Integration Points
With research-to-essay skill:
- Forge concept → Research empirical grounding → Write explanatory essay
With strategy-to-artifact skill:
- Forge concept → Create visual framework → Build presentation deck
With prose-polish skill:
- Ensure concept descriptions avoid generic AI language
- Polish final documentation
With user's voice signature (from research-to-essay):
- Use conversational transitions ("So," "But here's," "Hold on")
- Employ recursive refinement ("Let me be more precise")
- Include dialogue structure naturally
- Apply practitioner stance
Common Concept Types
Load references/archetype-voices.md for detailed paths and archetype pairings.
Common patterns: Taxonomic (classification grids), Process (maturity models), Causal (explanatory models), Diagnostic (decision heuristics), Constraint (strategic maps).
Example Triggers
- "I've been thinking about something but can't quite articulate it"
- "Explore this idea with me"
- "There's something about how AI changes coordination..."
- "Challenge my thinking on X"
- "Help me pressure-test this framework"
- "What if we thought about it as..."
- "I think X is actually Y, but not sure"
- "Walk me through why this matters"
Success Metrics
Concept forging succeeds when:
- User gains new clarity on previously vague intuition
- Structure emerges that wasn't visible before
- Concept is testable and falsifiable
- User can apply without further assistance
- Generates new questions or insights
- Different from existing concepts in meaningful way
Process succeeds when:
- User feels intellectually challenged (not just supported)
- Genuine dialectic (not Socratic theater)
- Archetype voices distinct and authentic
- Tensions acknowledged honestly
- User's thinking elevated (not just organized)
