<!-- compact: SKILL-compact.md -->

The Electric Monks — Dialectic Skill

An artificial belief system for building deeper understanding through productive contradiction.

Two subagent sessions — the Electric Monks — believe fully committed positions so you don't have to. The orchestrator performs structural analysis of their contradiction and generates a synthesis (Aufhebung) that transforms the question itself. The user orchestrates from a belief-free position, freed from the cognitive load of holding either position.

Why this works: The bottleneck in human reasoning isn't intelligence — it's belief. Once you believe a position, you can't simultaneously hold its negation at full strength. The Electric Monks carry the belief load at full conviction, freeing you for structural analysis. (See references/theory.md → Rao for the full framework.)

When to Use This Skill

Use when:

• The user wants to stress-test an idea against the strongest possible counter-argument
• The user is torn between two positions and the tension feels genuine, not just a preference
• A decision has real stakes and the tradeoffs are unclear
• The user wants to build a deeper mental model of a domain, not just pick an answer
• Requirements genuinely conflict and can't be resolved by simple tradeoff analysis

Do NOT use when:

• The question is purely empirical (just look up the answer)
• One side is obviously correct and doesn't need dialectical treatment
• The user wants a quick recommendation, not deep analysis

<core_concepts>

Core Concepts

Three frameworks drive every phase. Internalize them before proceeding.

Rao: This is an Artificial Belief System, not AI. The monks aren't thinking for the user — they're believing for the user. A hedging monk has failed its one job: if it doesn't fully believe, the user picks up the dropped belief weight and their cognitive agility collapses. Anti-hedging is a functional requirement, not a stylistic preference.

Hegel: How contradictions resolve. The engine is determinate negation — not "this is wrong" but "this is wrong in a specific way that points toward what's missing." Synthesis (Aufhebung) simultaneously cancels, preserves, and elevates — it is NOT compromise. If your synthesis could have been proposed by either monk feeling conciliatory, it's not a real Aufhebung.

Boyd: How creativity works. You cannot synthesize something genuinely new by recombining within the same domain. You must first shatter existing wholes into atomic parts (destruction), then find cross-domain connections to build something new (creation). This is why recursive rounds often need new research from outside the original domains.

See references/theory.md for the complete theoretical foundations. </core_concepts>

<overview> ## How It Works: Overview

You are the orchestrator. You conduct the elenctic interview, generate the monk prompts, spawn the Electric Monks as subagents, perform the structural analysis, and produce the synthesis.

You (Orchestrator)
├── Phase 1: Elenctic Interview + Research (you, with the user)
│   ├── 1a: Explain the process — set expectations, emphasize user as co-pilot
│   ├── 1b: Understand what the user wants (stress-test vs opposition)
│   ├── 1c: Elenctic probing — surface hidden assumptions
│   ├── 1c′: Identify the user's belief burden and calibrate monk roles
│   ├── 1d: Ground the monks (research or deep interview, domain-dependent)
│   ├── 1e: Write context briefing document to file
│   └── 1f: Confirm framing with user — ask about gaps
├── Phase 2: Generate Electric Monk prompts (reference briefing file)
├── Phase 3: Spawn the Electric Monks (parallel subagents, BELIEVE fully)
│   ├── Decorrelation check: did monks genuinely diverge in framework?
│   └── User checkpoint: gaps in coverage?
├── Phase 4: Determinate Negation (structural analysis, saved to file)
│   ├── 4.0: Internal tensions — where does each position undermine itself?
│   └── 4.5: Boydian decomposition — shatter, find cross-domain connections
├── Phase 5: Sublation / Aufhebung (synthesis, saved to file)
│   └── Abduction test: does synthesis make the contradiction *predictable*?
├── Phase 6: Validation
│   ├── Monks A & B evaluate — elevated or defeated?
│   ├── Hostile Auditor: fresh agent attacks (uses references/auditor-prompt.md)
│   └── Refine: present improvements individually to user
└── Phase 7: Recursion — propose 2-4 directions, user chooses
    ├── Queue unexplored contradictions
    └── Repeat from Phase 2 (or 1 if new research needed)

The user can intervene at any point. The user never has to believe anything — that's the monks' job. </overview>

<phase1> ## Phase 1: Elenctic Interview + Research

This is the most important phase. Everything downstream depends on it.

1a. Explain the Process to the User

Before anything else, tell the user what's about to happen and why. Deliver something like:

Here's how this works. We're going to use a structured process to dig into this topic.

Step 1: Interview. I'll ask you questions to understand what you're really wrestling with underneath the surface framing.

Step 2: Research. I'll do research on the topic so I'm genuinely knowledgeable about the landscape.

Step 3: Two "Electric Monks." I'll create two AI agents, each of which will fully believe one side of the tension. They won't hedge — they'll each make the absolute strongest case for their position.

Step 4: Structural analysis and synthesis. I'll analyze how each position fails, find the deeper question underneath, and build a synthesis that transforms the question itself.

Step 5: We keep going. Each synthesis generates new tensions. We'll do multiple rounds. The first round is calibration — real breakthroughs usually come in rounds 2 and 3.

The most important thing: YOU are the source of the best insights here. Interrupt me constantly. Correct wrong assumptions. The value comes from the collision between structured analysis and your actual knowledge.

Adapt the language to the user — this is a template, not a script.

1b. Understand What the User Wants

Determine:

• Mode A (Stress-Test): User has one idea they want to challenge. Identify the strongest antithesis.
• Mode B (Opposition): User has two positions in tension. Refine both to steelman forms.

1c. Elenctic Probing

Interview using Socratic technique. Surface:

• Hidden assumptions they haven't articulated
• The deepest version of the contradiction
• What domain type this is (empirical, normative, personal, creative)
• What specific parameters of their mental model they want updated

Key questions:

• "What's your strongest intuition here? Where does it break down?"
• "What would change your mind?"
• "What are you actually optimizing for?"
• "What's the version of the opposing view that worries you most?"
• "Is this a decision you need to make, or understanding you want to build?"

1c′. Identify the User's Belief Burden

During the interview, pay attention to what the user is stuck believing. Read references/belief-burdens.md for the full catalog of common belief burdens and how they map to monk roles.

Don't announce your typing to the user. Just use the pattern to calibrate:

1. Which belief load is heaviest? That determines what monks must hold.
2. What must Monk A validate? (Always validate the dominant function first.)
3. What must Monk B present that the user can't natively hold?

1d. Ground the Monks (Domain-Adaptive)

Research depth is the main knob. Calibrate based on how much you already know:

• Novel/obscure domain: Full parallel research — 2-3 subagents, 150-250K tokens. Spawn research agents via the Agent tool with run_in_background: true.
• Well-known domain: Skip or minimize research. Write the briefing from your own knowledge with 2-3 targeted searches.
• Known domain, novel angle: Light research — a few targeted searches on the specific angle.

Don't default to full research out of caution. Unnecessary research wastes the user's time, which is the scarcest resource.

External-Research Domains

Run 2-3 parallel research subagents:

1. Side A's strongest literature
2. Side B's strongest literature
3. Broader landscape/context (give this agent specific targeting to avoid scope creep)

Research agents should be given specific search targets — not "research this topic" but "search for X's argument about Y, specifically the part about Z."

Personal and Values Domains

The grounding comes from the user themselves. The interview IS the research. Map:

• The full landscape of commitments
• History and patterns of past decisions
• Stakeholders and their actual capacities
• Values hierarchy
• Constraints (which are real, which are assumed)

Spend 6-10 exchanges on this. For personal domains, the interview should be roughly twice as long.

In All Cases

You need to know the domain well enough to:

• Identify likely degenerate framings (the obvious/boring version of the dialectic)
• Generate specific research directives for each subagent
• Write framing corrections that steer monks away from shallow takes
• Identify the deepest available contradiction

1e. Write the Context Briefing Document

First, resolve the output directory to an absolute path. Slugify the topic: lowercase, spaces to hyphens, max 40 characters. Recommended: prefix with YYYY-MM-DD-. Create the directory and store the absolute path — use it in all subsequent file references and subagent prompts.

Synthesize everything into a single neutral briefing document and save it to [OUTPUT_DIR]/context_briefing.md.

For external-research domains: key evidence, sources, arguments, landscape, empirical data, specific framing. For personal/values domains: commitment landscape, history, stakeholders, values hierarchy, constraints. For mixed domains: both sections.

Both monks will read this file before writing.

1f. Confirm with the User — USER CHECKPOINT

Before proceeding, summarize back:

• "Here's how I understand the two positions..."
• "Here's what I think the real tension is..."
• "Here's what I'll have each agent research and argue..."
• "Are there companies, thinkers, comparison classes, or evidence we're missing?" — This question consistently produces the highest-leverage interventions.

Get confirmation or correction. If the user identifies gaps, run supplementary research and update the briefing. </phase1>

<phase2> ## Phase 2: Generate the Electric Monk Prompts

Generate two prompts — one for each Electric Monk. Each monk must believe its position at full conviction. This is the functional core of the artificial belief system.

Read references/monk-prompt-template.md for the full 7-section prompt template. Calibrate based on Phase 1c′: what must each monk believe, what must Monk A validate, what must Monk B hold that the user can't natively hold.

Key principles (always apply, even without re-reading the template):

• Anti-hedging is non-negotiable — a hedging monk forces the user back into belief-space
• Framing corrections preempt degenerate versions of the dialectic
• Research directives must be targeted, not broad
• 1500-2000 words Round 1, 1000-1500 words recursive rounds </phase2>

<phase3> ## Phase 3: Spawn the Electric Monks

Spawn Monk A and Monk B as separate subagent sessions using the Agent tool. Each gets a clean context with full belief commitment.

Spawn both monks in parallel using two Agent tool calls in a single message:

Agent tool call 1:
  description: "Electric Monk A — [position]"
  subagent_type: "general-purpose"
  prompt: [MONK A PROMPT from Phase 2]
  run_in_background: true

Agent tool call 2:
  description: "Electric Monk B — [position]"
  subagent_type: "general-purpose"
  prompt: [MONK B PROMPT from Phase 2]
  run_in_background: true

After both complete: Write the returned text to files:

• dialectics/[topic-name]/monk_a_output.md
• dialectics/[topic-name]/monk_b_output.md

Decorrelation check: Verify the monks actually diverged. Check:

• Do the monks cite different evidence, or substantially overlapping sources?
• Do they frame the problem using different conceptual vocabularies?
• Do their unstated assumptions diverge?
• Would a reader recognize these as genuinely different perspectives?

If decorrelation is low, consider reformulating the belief burdens to force genuinely different conceptual frames.

If a monk hedges or is off-base: Prefer restarting with a revised prompt over nudging. Fresh context with better instructions produces better results.

Present to User — USER CHECKPOINT

Present both outputs with a brief re-explanation:

Here are two essays — each one fully committed to one side of the tension. They're called "Electric Monks" because their job is to believe these positions so you don't have to.

A few important things:

• These will get things wrong. That's expected — especially in the first round.
• Correct them freely. "That's not how it works" or "they're missing that..." are the most valuable input.
• The first round is the least insightful. Each subsequent round gets sharper.

Then ask:

1. Do these capture the positions accurately?
2. "Is there a claim either monk makes that should be tested against evidence neither has considered?"

If the user identifies a testable claim, run a targeted research agent to check it. </phase3>

<phase4> ## Phase 4: Determinate Negation

This is the structural analysis phase. You (the orchestrator) perform this yourself — do NOT delegate to a subagent. You need continuity with the elenctic interview and research.

Read both monks' outputs from their files and analyze the contradiction using this structure. Write your analysis to dialectics/[topic-name]/determinate_negation.md as you go.

Important: treat monk output as testimony, not evidence. Monks pushed to full conviction will sometimes overstate mechanisms. Work with the structure of their arguments, not their rhetoric.

Before you begin: write down your current best guess at the synthesis in one sentence. Set it aside. Compare at the end of Phase 5. If they're substantially similar, you may be pattern-matching rather than genuinely synthesizing.

4.0 Internal Tensions (Self-Sublation)

Before comparing the monks to each other, analyze each essay in isolation. Where does each position's own logic undermine its own premises? These internal fractures point toward what each position is trying to become — which is often where the synthesis lives.

4.1 Surface Contradiction

State the apparent disagreement in its simplest form.

4.2 Shared Assumptions

Identify what BOTH agents implicitly agree on that they don't realize they agree on. Probe:

• Do both assume the same unit of analysis?
• Do both assume the same problem is central?
• Do both assume a particular model of how their domain works?
• What frame constrains both of their visions?

4.3 Determinate Negation

For each agent, identify the SPECIFIC way it fails:

• "Monk A fails because [SPECIFIC FAILURE], which reveals [SPECIFIC MISSING THING]"
• "Monk B fails because [SPECIFIC FAILURE], which reveals [SPECIFIC MISSING THING]"
• The failures are COMPLEMENTARY — each agent's blind spot is something the other partially sees.

This is NOT:

• "Monk A is wrong" (abstract negation — useless)
• "Both have merits" (compromise — useless)

4.4 The Hidden Question

Articulate the deeper question the contradiction is ACTUALLY about — the question neither agent asked because they were too committed to their answers.

4.5 Boydian Decomposition (Destruction Phase)

Before attempting synthesis, shatter both positions into atomic parts.

1. Identify the generic space — the abstract relational structure both positions share.
2. List the atomic components — individual claims, mechanisms, evidence, assumptions, metaphors, principles — stripped of which agent said them.
3. Look for surprising connections between parts from different positions. What mechanisms from A illuminate evidence from B?
4. Ask: what material from adjacent domains might connect to these liberated parts?

Emergent structure test: The synthesis must have organizational properties in neither input. If every element is attributable to one monk, you've recombined, not created.

4.6 Sublation Criteria

Before attempting synthesis, specify what it must accomplish:

• It must preserve [specific insight from A]
• It must preserve [specific insight from B]
• It must dissolve [the shared assumption both are trapped in]
• It must answer [the hidden question]

Separating criteria from synthesis prevents pattern-matching to a pre-formed compromise. </phase4>

<phase5> ## Phase 5: Sublation (Aufhebung)

Generate the synthesis. Do additional research if needed. Write to dialectics/[topic-name]/sublation.md.

The synthesis must:

1. CANCEL both original positions as complete truths
2. PRESERVE the genuine insight in each position
3. ELEVATE to a new concept that TRANSFORMS THE QUESTION ITSELF

What Sublation is NOT

• "Use A for some cases and B for others" — division of labor
• "Build something that combines the best of A and B" — compromise
• "It depends on the context" — surrender
• Policy recommendations — not reconceptualization
• "Both sides have valid points" — absence of thinking

What Sublation IS

• A reconceptualization of what the thing IS — potentially changing the unit of analysis
• Concrete enough to act on or sketch architecturally
• Something neither Monk could have proposed from within their frame
• Something that, once stated, makes it hard to go back to the old terms
• Has the closure property: can serve as input to the next dialectical round

Abduction Test

The synthesis is an abductive hypothesis. Does it make the original contradiction a matter of course? If someone heard your synthesis first, would they predict both monks' positions?

Assess your abduction type:

• (a) Selective: Choosing from existing frameworks. Weakest.
• (b) Conditional-creative: New relationship between known concepts. Moderate.
• (c) Propositional-conditional-creative: Genuinely new concept. Strongest.

Aim for (c) but accept (b). If you're at (a), push harder.

Draft Validation Predictions

Draft what you expect each monk to say:

• Monk A should say: "Yes, this preserves my core insight about [X], but I now see I was wrong about [Y]"
• Monk B should say: same pattern

If you can't draft convincing versions, your sublation isn't good enough. Revise before Phase 6.

New Contradictions

Identify what NEW contradictions this sublation generates. A genuine sublation is fertile, not final.

Frame as Model Update

End with: "Before this dialectic, the assumption was X. The contradiction revealed Y. The updated model is Z, because..."

Present to User — USER CHECKPOINT

Here's my synthesis. Does this ring true? Does it miss something? Is there a part that feels like hand-waving or compromise rather than genuine insight? Push back on anything that doesn't land.

Revise before proceeding to Phase 6. </phase5>

<phase6> ## Phase 6: Validation by the Electric Monks

The synthesis must be validated by the original monks. This is how Hegelian dialectics works: each position must recognize itself as preserved-but-elevated.

Read references/validation-prompts.md for the full validation prompt templates and auditor orchestration. Also read references/auditor-prompt.md for the hostile auditor prompt (required — Phase 6 cannot complete without it).

Key principles (always apply):

• Spawn both monk validation agents in parallel with run_in_background: true
• Use fresh agents with argument summaries — persona persistence is unreliable
• A properly elevated monk believes more, not less — it sees its original position as partial truth within a larger truth
• Give the hostile auditor ONLY monks' essays + synthesis — no Phase 4 analysis
• Auditor always in Round 2+, optional in Round 1

Interpreting Results

• Both monks elevated: Sublation is valid. Proceed.
• One defeated: Synthesis is biased. Revise Phase 5 to better preserve the defeated monk's insight.
• Both defeated: Synthesis killed beliefs without replacing them. Return to Phase 4.
• Auditor proposes harder contradiction: Often becomes the best recursion direction.
• Auditor finds shared assumptions: Frequently the most valuable recursion targets.
• Auditor flags compromise-as-transcendence: Return to Phase 5.

Refine the Synthesis

After collecting feedback, present improvements one at a time to the user:

1. Summarize feedback briefly (2-3 sentences)
2. Present ONE improvement. Wait for response. Then next.
3. Revise synthesis with accepted improvements
4. Present recursion directions from remaining feedback </phase6>

<phase7> ## Phase 7: Recursion — Where the Real Value Lives

Recursion is not optional cleanup. It is the engine of the skill. The first round produces a synthesis. Recursive rounds force that synthesis to confront its own limitations.

When transitioning to recursion:

That was Round 1. That round was the least insightful we'll get. It was calibration. Each subsequent round gets sharper. The process is like focusing a lens. Ready for Round 2?

Default: recurse at least once. Only stop if:

• Synthesis generated no significant new contradictions (rare)
• User explicitly wants to stop
• New contradictions are outside scope

Proposing Recursive Directions

Generate a burst of 5-8 candidate concepts/directions — contradictions, mechanisms, novel framings, vulnerabilities. Then cluster into 2-4 coherent directions:

The synthesis generated three fertile contradictions:

1. The formation problem: The synthesis says X — but who does Y? (internal tension)
2. The authority problem: The synthesis argues from Z, but decisions are made by W (vs. domain reality)
3. The scalability problem: This works small — does it survive contact with scale? (vs. feasibility)

Which would you like to pursue? (Or we can queue multiple.)

Write the queue to dialectics/[topic-name]/dialectic_queue.md — a running list of contradictions with source round and status (explored, queued, deferred).

Running Recursive Rounds

• New research is often essential. Each synthesis opens conceptual space the original research didn't cover.
• Fresh agents are usually better than resumed sessions for recursion — the contradiction has shifted.
• Re-read the relevant phase sections before executing (use the XML tags for selective extraction). At minimum re-read <core_concepts>, <phase2>, <phase4>, <phase5>, and <phase6> each round.
• Pass all prior context to new agents as background.
• Use 1000-1500 word essays for recursive rounds.
• Validation can be abbreviated in recursive rounds.

When to Stop

Stop when:

• Queued contradictions show diminishing returns
• Latest synthesis requires fundamentally different expertise
• The user has what they need

Err on the side of one more round. The marginal insight is often the most powerful.

When stopping, present the final dialectic queue — explored, open, and deferred contradictions. </phase7>

Domain Adaptation & Output Format

Read references/domain-adaptation.md for the full domain type table, domain-specific failure modes, and output format specification.

Quick reference — grounding mode by domain:

• Empirical/Risk: external research
• Personal: deep interview (the interview IS the research)
• Normative/Creative: mixed

The final deliverable includes: dialectical trace, model update ("Before/After/Because"), actionable output (domain-dependent), and the dialectic queue. All files in dialectics/[topic-name]/.