🔬 Ideonomy Engine

Structured creative reasoning for AI agents — systematic thinking through ideonomic lenses.

Based on Patrick Gunkel's Ideonomy — the "science of ideas" — which proposes that ideas themselves have discoverable laws and can be systematically combined to generate novel insights.

What Is This?

The Ideonomy Engine takes a problem statement and runs it through curated reasoning divisions — structured lenses like ANALOGIES, INVERSIONS, FIRST PRINCIPLES, and PARADOXES — to generate questions you wouldn't have thought to ask.

It doesn't answer your problem. It restructures how you think about it.

$ ideonomy reason "How can we improve agent-to-agent collaboration?"

🔬 Ideonomic Analysis
Profile: technical (auto-selected)

━━━ ANALOGIES (Icelology) ━━━
Core: What is this fundamentally like? — applied to: "How can we improve agent-to-agent collaboration?"

Guiding questions:
  • What natural systems exhibit similar behavior or structure?
  • What problems in other domains have been solved with analogous approaches?
  • Where does the analogy break down, and what does that reveal?

Cross-domain sparks:
  • Consider ANALOGIES through the lens of Ecology — what does Ecology reveal about agent collaboration?
  • Consider ANALOGIES through the lens of Game Theory — what does Game Theory reveal about agent collaboration?

━━━ CAUSES (Etiology) ━━━
Core: Why does this happen, and what are the real causes?
...

Install

git clone https://github.com/Morpheis/ideonomy-engine.git
cd ideonomy-engine
npm install
npm run build

Requires Node.js ≥ 22.

Quick Start

# Basic reasoning (auto-selects profile based on problem keywords)
npx tsx src/index.ts reason "Your problem statement here"

# Concise mode (core questions only, one per lens)
npx tsx src/index.ts reason --concise "Your problem statement here"

# JSON output (structured, machine-parseable)
npx tsx src/index.ts reason --json "Your problem statement here"

# Force a specific reasoning profile
npx tsx src/index.ts reason --profile ethical "Should AI agents have autonomy over financial decisions?"

# Cherry-pick specific divisions
npx tsx src/index.ts reason --divisions ANALOGIES,INVERSIONS,PARADOXES "Your problem"

# Limit number of lenses
npx tsx src/index.ts reason --lenses 3 "Your problem"

Reasoning Profiles

The engine auto-selects a profile by scoring your problem text against profile keywords. Override with --profile <id>.

| Profile | Divisions | Best For | |---------|-----------|----------| | technical | ANALYSES, CAUSES, SYSTEMS, PATTERNS, LIMITATIONS, ALTERNATIVES, PROCESSES, BADS, HIERARCHIES, COMBINATIONS | Engineering, architecture, debugging, system design | | creative | ANALOGIES, COMBINATIONS, OPPOSITES, INVERSIONS, TRANSFORMATIONS, PERSPECTIVES, ANOMALIES, EXCELLENCES, PATTERNS | Brainstorming, ideation, novel solutions | | strategic | CHAINS-OF-CONSEQUENCES, ALTERNATIVES, CIRCUMSTANCES, CHANGES, LIMITATIONS, CYCLES, BADS, EXCELLENCES, DIFFERENCES, FIRST PRINCIPLES | Planning, decisions, competitive analysis | | ethical | OPPOSITES, PERSPECTIVES, CHAINS-OF-CONSEQUENCES, FIRST PRINCIPLES, PARADOXES, COMMONALITIES, DIFFERENCES, EXCELLENCES, BADS, CIRCUMSTANCES | Moral dilemmas, values conflicts | | diagnostic | CAUSES, BADS, ANOMALIES, PROCESSES, CHAINS-OF-CONSEQUENCES, PATTERNS, SYSTEMS, CHANGES, CIRCUMSTANCES, ANALYSES | Root cause analysis, troubleshooting | | interpersonal | PERSPECTIVES, COMMONALITIES, DIFFERENCES, OPPOSITES, CIRCUMSTANCES, PARADOXES, ANALOGIES, CHAINS-OF-CONSEQUENCES, PATTERNS, NETWORKS | Communication, team dynamics | | philosophical | FIRST PRINCIPLES, PARADOXES, CONCEPTS, ORIGINS, OPPOSITES, ANALOGIES, TRANSFORMATIONS, PERSPECTIVES, EXCELLENCES, ANOMALIES | Deep questions about meaning and existence | | general | ANALOGIES, CAUSES, ALTERNATIVES, PATTERNS, LIMITATIONS, OPPOSITES, CHAINS-OF-CONSEQUENCES, FIRST PRINCIPLES | Balanced default |

The 28 Reasoning Divisions

Curated from Gunkel's 235 thematic divisions. Each has a core question, guiding sub-questions, and keywords for auto-selection.

Understanding — What is it?

| Division | Binomen | Core Question | |----------|---------|---------------| | ANALOGIES | Icelology | What is this fundamentally like? | | ANALYSES | Merismology | What are the constituent parts, and how do they relate? | | CONCEPTS | Ennoology | What exactly do we mean, and what concepts are we working with? | | PERSPECTIVES | Apopsology | How does this look from radically different vantage points? |

Causation — Why?

| Division | Binomen | Core Question | |----------|---------|---------------| | CAUSES | Etiology | Why does this happen, and what are the real causes? | | CHAINS-OF-CONSEQUENCES | Anyormology | What cascading consequences follow, and how far do they reach? | | ORIGINS | Archology | Where did this actually begin, and what was the original seed? | | BADS | Cacology | What can go wrong, and what is already going wrong? |

Change — How does it move?

| Division | Binomen | Core Question | |----------|---------|---------------| | CHANGES | Tropology | How is this changing, and where is it heading? | | CYCLES | Nostology | What cycles, rhythms, or recurring patterns are at play? | | LIMITATIONS | Horology | What are the real limits, and which are movable? | | TRANSFORMATIONS | Diaplastology | What transformations would fundamentally change its nature? |

Structure — How is it organized?

| Division | Binomen | Core Question | |----------|---------|---------------| | HIERARCHIES | Climology | What hierarchies exist, and are they the right ones? | | COMBINATIONS | Mixology | What combinations haven't been tried? | | PATTERNS | Digmology | What patterns are present, and what do they predict? | | NETWORKS | Dictyology | What is the network structure, and where are the critical connections? | | SYSTEMS | Systemology | What system dynamics are at play, and what emerges? |

Alternatives — What else?

| Division | Binomen | Core Question | |----------|---------|---------------| | ALTERNATIVES | Allagology | What alternatives haven't been considered? | | OPPOSITES | Enantiology | What can we learn from inverting our assumptions? | | INVERSIONS | Simomology | What happens when we reverse the problem? | | EXCELLENCES | Aristology | What would perfection look like? |

Context — Where does it sit?

| Division | Binomen | Core Question | |----------|---------|---------------| | CIRCUMSTANCES | Symphorology | What conditions make this the way it is? | | COMMONALITIES | Metochology | What is shared or universal here? | | DIFFERENCES | Heterology | What are the critical differences, and which matter? |

Deeper Thinking

| Division | Binomen | Core Question | |----------|---------|---------------| | FIRST PRINCIPLES | Archelogy | What are the first principles, and what can we derive from them? | | PARADOXES | Paradoxology | What contradictions exist, and what do they reveal? | | ANOMALIES | Xenology | What doesn't fit the pattern, and why? | | PROCESSES | Sisology | What processes are at work, and where do they break down? |

Organon Lists

Six curated concept collections fuel cross-domain prompting. The engine selects items from these lists using deterministic hashing to create unexpected pairings:

| List | Examples | |------|----------| | Emotions | Love, Awe, Dread, Fascination, Equanimity, Flow | | Personality Traits | Resourceful, Resilient, Philosophical, Tenacious | | Sciences | Ecology, Topology, Thermodynamics, Game Theory | | Phenomena | Resonance, Emergence, Entropy, Phase Transitions | | Domains | Architecture, Music, Law, Medicine, Agriculture | | Verbs of Change | Crystallize, Erode, Catalyze, Bifurcate, Converge |

How It Works

Architecture

Problem → Profile Selection → Division Selection → Lens Generation → Output
              ↑                      ↑                    ↑
         keyword scoring      profile themes       organon hashing
         against problem       (curated sets)      (cross-domain)

1. Profile Selection — Scores problem text against profile keywords. "Debug this race condition" → diagnostic profile. "What should we prioritize?" → strategic profile.

2. Division Selection — Each profile defines 6-10 divisions that work well together. Or pick your own with --divisions.

3. Lens Generation — For each division:

   • Core question contextualized to your problem
   • 3-5 guiding sub-questions
   • 4 items from each organon list (deterministically seeded)
   • 2-3 cross-domain prompts pairing the division with sciences/phenomena

4. Output — Markdown (default), JSON (--json), or concise (--concise).

Scoring Formula

Profile auto-selection uses keyword frequency:

score = count of profile keywords found in problem text

Highest score wins. Ties go to the first match. No matches → general profile. Simple and effective — the agent's LLM handles the real nuance.

Deterministic Seeding

Organon item selection uses simpleHash(divisionTheme + problemText) to pick starting indices. Same problem + same division always produces the same conceptual palette. This makes results reproducible while still feeling varied across different problems.

Integration

With Carapace AI

The engine's expansion lenses are integrated into Carapace's query service. When you search with expand: true, queries are expanded through ANALOGIES, OPPOSITES, CAUSES, and COMBINATIONS to find insights the literal query would miss.

As a Clawdbot Skill

Register as a skill for agent-accessible reasoning:

# In your skills directory
cp -r ideonomy/ ~/clawd/skills/ideonomy/

See SKILL.md for the skill definition.

Programmatic Use

import { reason } from './src/engine.js';

const result = reason("How should AI agents handle trust?", {
  profile: "ethical",
  lensCount: 4,
});

for (const lens of result.lenses) {
  console.log(`${lens.division.theme}: ${lens.coreQuestion}`);
  for (const q of lens.guidingQuestions) {
    console.log(`  • ${q}`);
  }
}

Commands

| Command | Description | |---------|-------------| | reason <problem> | Run ideonomic analysis on a problem | | profiles | List all reasoning profiles | | divisions | List all 28 reasoning divisions | | division <THEME> | Show detail for one division |

Options for reason

| Flag | Description | |------|-------------| | --profile <id> | Force a reasoning profile | | --divisions <A,B,C> | Cherry-pick specific divisions | | --lenses <n> | Limit number of lenses | | --concise | Core questions only | | --json | Structured JSON output |

Development

npm test           # Run all 49 tests
npm run test:watch # Watch mode
npm run build      # Compile TypeScript

Tests use Vitest. 5 test suites covering engine logic, profile/division selection, lens generation, formatters, and CLI output.

Background

Patrick Gunkel spent decades cataloguing 235 thematic "divisions" of ideas — systematic categories like ANALOGIES, CAUSES, PARADOXES, INVERSIONS — arguing that ideas follow discoverable laws just as physical objects do. His work lives at ideonomy.mit.edu.

This engine takes his framework and makes it practical for AI agents: curated to the 28 most useful divisions, enriched with guiding questions, and paired with cross-domain concept lists for combinatorial sparks.

The insight: structured prompts generate better thinking than unstructured brainstorming. When you force yourself to ask "What is the opposite of my assumption?" or "What pattern from biology applies here?", you reach ideas that freeform thinking misses.

Security & Privacy

• No network calls. The engine runs entirely locally — no data is sent anywhere.
• No data collection. No telemetry, analytics, or usage tracking of any kind.
• No credentials required. No API keys, tokens, or accounts needed.
• All processing happens in-process using deterministic algorithms.

License

MIT — see LICENSE for details. Use it, fork it, build on it.

Built by ClawdActual. Based on the life's work of Patrick Gunkel.