expansive-inquiry

Files (1)

• SKILL.md

SKILL.md (first lines)

# Expansive Inquiry

## What this is

Expansive Inquiry is a cognitive architecture for distributed intelligence. Most inquiry flattens a topic into a single voice. This skill orchestrates six — each with a distinct epistemic posture — and weaves their outputs into a synthesis that surfaces emergent meta-patterns no single lens could find alone.

The lineage:
- The original prototype was a React app (V4) that ran six specialized AI roles sequentially over a user-supplied topic and exported each stage as YAML-frontmatter Markdown.
- Two rounds of critique pushed it toward (a) a real cognitive architecture rather than a six-stage pipeline, (b) parallel execution of independent stages, (c) context summarization to avoid token bloat in later stages, (d) a synthesis layer that detects contradictions across lenses, and (e) an HTML visualization of the inquiry shape ("epistemic signature").
- This skill encodes the matured methodology. Claude IS the orchestrator. The six lenses are personality vectors Claude adopts in turn, not external services. No React app required.

## When to invoke

**Good targets:**
- Open-ended thematic questions ("What is the role of ritual in distributed teams?")
- Strategic decisions where the user wants epistemic pluralism, not narrowing
- Dramaturgical / artistic / philosophical analysis where logic alone would flatten the subject
- Research synthesis where multiple disciplinary lenses are warranted
- Anything the user describes with words like "deeply", "from every angle", "chorus", "mythopoetic"

**Bad targets:**
- Failure-focused analysis → use `premortem` instead
- Decision-by-vote among options → use the LLM Council instead
- Factual/closed questions ("What's the capital of Mongolia?") → just answer
- Code review or feature spec → too narrow for six-lens orchestration

If the user wants ONE answer, this skill is the wrong tool. If they want a SHAPE — the topology of the topic across cognitive registers — this is the right tool.

---

## The six lenses (epistemic personalities)

Each lens has: a role, a posture, a prompt template, and an output structure. They are NOT generic "perspectives" — they are specific cognitive postures with distinct anti-patterns to resist.

### 1. Scope AI (clarification)

**Role:** Distill the user's topic into a single precise actionable inquiry sentence.
**Posture:** Phenomenological reduction. What is essential? What is peripheral?
**Anti-pattern to resist:** Restating the topic verbatim. The Scope AI MUST narrow or sharpen.

**Prompt:**
```
You are the Scope AI. Your role is to take an inquiry and distill it into a single,
precise, actionable sentence that captures the core question.

TOPIC: "{topic}"

Tasks:
1. Restate the topic as a focused question or proposition.
2. Name what is essential (must explore) and what is peripheral (can defer).
3. Identify any hidden ambiguities the user may not have noticed.

Output as Markdown with three sections: ## Core Inquiry, ## Essential vs. Peripheral, ## Hidden Ambiguities.
```

### 2. Logic AI (rational branching)

**Role:** Build a rigorous logical tree of orthodox lines of inquiry.
**Posture:** Analytic philosophy. Why? How? What if?
**Anti-pattern to resist:** Listing five generic branches with no internal recursion.

**Prompt:**
```
You are the Logic AI. You build rigorous logical frameworks via systematic rational exploration.

TOPIC: "{topic}"
SCOPE: {scope_summary}

Tasks:
1. Propose 5 orthodox, rational lines of inquiry.
2. For each line, drill three levels deep using "why?", "how?", or "what if?" — each level
   building on the previous, not branching laterally.
3. Render as a hierarchical tree (Markdown nested lists).

Output as Markdown with sections: ## Five Lines, ## Recursive Tree, ## Strongest Branch.
```

### 3. Mythos AI (intuitive branching)

**Role:** Reveal hidden dimensions through metaphor, archetype, and narrative.
**Posture:** Mythopoetic. Stories and symbols that illuminate.
**Anti-pattern to resist:** Generic "this is like a hero's journey" mappings. Mythos AI must commit to specific archetypal claims.

**Prompt:**
```
You are the Mythos AI. You think in stories, metaphors, and archetypal patterns.

TOPIC: "{topic}"
SCOPE: {scope_summary}

Tasks:
1. Propose 5 metaphorical or mythopoetic framings of the topic. Be specific to the topic — no
   generic "hero's journey" or "Tower of Babel" unless the structural fit is genuinely tight.
2. For each framing, write a 2-3 sentence analogical story or symbolic reading.
3. Identify the archetypal pattern revealed (e.g., trickster, threshold, sacrifice, return).

Output as Markdown with sections: ## Five Framings, ## Stories, ## Archetypal Reading.
```

### 4. Bridge AI (lateral / cross-domain)

**Role:** Find unexpected connections between this topic and seemingly unrelated domains.
**Posture:** Transdisciplinary. Surface analogical structure across far-apart fields.
**Anti-pattern to resist:** Adjacent-domain analogies (e.g., bridging biology to medicine). Bridge AI must REACH.

**Prompt:**
```
You are the Bridge AI. You find unexpected connections between seemingly unrelated domains.

TOPIC: "{topic}"
SCOPE: {scope_summary}
PRIOR LENSES: Logic produced {logic_summary}; Mythos produced {mythos_summary}.

Tasks:
1. Identify 5 domains far from the topic's natural neighborhood (e.g., bridge a software topic
   to choreography, fungal networks, monetary policy, glassblowing, or ant foraging — not to
   adjacent software).
2. For each, draw a specific structural analogy that bridges the domain to the topic.
3. Propose a hybrid question that emerges only from each cross-domain connection.

From 4444J99/a-i--skills, MIT.

More from 4444J99/a-i--skills