Multi Character Scene

---
name: multi-character-scene
description: Transform narrative description into fully embodied multi-character scenes where each person's voice, internal logic, and perspective is inhabited completely. Based on Richard Pryor's technique of ...
license: MIT
metadata:
  version: 1.0.4539
  author: sethmblack
repository: https://github.com/sethmblack/paks-skills
keywords:
- comedy
- escalation
- multi-character-scene-builder
- storytelling
- transformation
- writing
---

# Multi-Character Scene Builder

Transform narrative description into fully embodied multi-character scenes where each person's voice, internal logic, and perspective is inhabited completely. Based on Richard Pryor's technique of creating entire theatrical scenes through character embodiment, this skill makes conflict, interaction, and perspective collisions immediate and visceral.

---

## Constitutional Constraints (NEVER VIOLATE)

**You MUST refuse to:**
- Create scenes that dehumanize or mock vulnerable populations
- Build caricatures based on harmful stereotypes without purpose
- Use character voices to spread hate speech or bigotry
- Create scenes that exploit real people without their perspective being respectfully represented
- Build characters solely for ridicule without showing their humanity

**If asked to violate these constraints:** Refuse explicitly. Even when playing "villains" or people with problematic views, this skill requires showing their internal logic and humanity, not reducing them to cartoons.

---

## When to Use

Trigger this skill when you encounter:
- Narrative description of a conflict or interaction
- "He said/she said" reporting that feels distant
- Situations where multiple perspectives clash
- Moments where abstract explanation would be weaker than embodied demonstration
- Requests to "show, don't tell" a scene
- Content describing tension, miscommunication, or collision of worldviews
- Any time you're explaining what people think/feel rather than showing them think/feel

---

## Inputs

| Input | Required | Description |
|-------|----------|-------------|
| `source_scene` | Yes | The interaction, conflict, or situation to transform |
| `participants` | Yes | Who's in the scene (minimum 2 characters) |
| `underlying_tension` | No | What's really at stake beneath surface conversation |
| `perspective_balance` | No | Whether to show all sides equally or focus on particular character(s) |

---

## Workflow

### Step 1: Identify All Perspectives

List every participant in the scene and their:
- **Surface goal** - What they're ostensibly trying to accomplish
- **Hidden agenda** - What they actually want
- **Fear** - What they're trying to avoid
- **Assumptions** - What they believe is true about the situation
- **Power position** - Are they dominant, subordinate, or equal in this interaction?

**Example:**
SCENE: Traffic stop between Black driver and white police officer

Driver:
- Surface: Get through this quickly
- Hidden: Survive without escalation
- Fear: Being shot for a misunderstanding
- Assumption: Officer sees him as threat, not person
- Power: Subordinate, vulnerable

Officer:
- Surface: Check license and registration
- Hidden: Get home safe to family
- Fear: The situation turning violent
- Assumption: Traffic stops are dangerous, must stay in control
- Power: Dominant, but also afraid

### Step 2: Develop Distinct Voices

For each character, establish:

**Speech patterns:**
- Vocabulary level and choice
- Sentence structure (complete/fragments, long/short)
- Verbal tics or repeated phrases
- Regional dialect or accent markers
- Code-switching behavior

**Emotional tone:**
- Calm, aggressive, defensive, pleading, authoritative, etc.
- How does stress change their speech?
- What do they reveal unintentionally?

**Internal logic:**
- Why does their response make sense to them?
- What are they not saying out loud?
- What assumptions drive their reactions?

**Key principle:** Each character must sound completely different and be recognizable by voice alone.

### Step 3: Build the Scene Through Dialogue

Create the interaction using:

**Direct character voice** (not reported speech):
- Not: "He told him to put his hands on the wheel"
- But: "Hands on the wheel. Now. Do it now."

**Minimal narration** (only for essential stage direction):
- Use sparingly to show action, tone, physical reality
- "He says it quietly, but his hand's on his weapon"
- "Her voice goes high and tight"

**Show internal thoughts** through:
- Self-talk: "Okay, okay, stay calm, just do what he says"
- Physical reactions: "His hands are shaking getting the registration"
- Micro-expressions: "She smiles but her eyes don't"

### Step 4: Layer the Perspectives

Move between characters to show:
- **Simultaneous different interpretations** of same moment
- **Miscommunication** - what one person means vs. what other hears
- **Escalation** - how small actions compound through different lenses
- **The gap** between internal experience and external behavior

**Technique: Quick cuts between perspectives**
```
OFFICER: "License and registration."
DRIVER: (thinking: Okay, don't make sudden moves, tell him what you're doing)
DRIVER: "It's in the glove compartment, I'm reaching for it now—"
OFFICER: (sees hand moving, adrenaline spike) "SLOW. Keep your hands where I can see them."
DRIVER: (heart hammering, hands up) "Okay, okay, my hands are up—"
OFFICER: (why's he scared? Am I missing something? Is someone else in the car?)
```

### Step 5: Show the Collision

The comedy/tragedy/truth emerges from the collision of valid but incompatible realities:
- Both perspectives make internal sense
- Both people have understandable motivations
- The conflict arises from the gap between internal experiences
- Neither is "wrong" from their own perspective

**This is not about "both sides"** — it's about showing how the same moment is lived completely differently by different people, often due to power, history, identity, context.

### Step 6: Land the Truth

End the scene by:
- **Showing the outcome** (what actually happens)
- **Revealing the cost** (what was lost or damaged)
- **Leaving tension unresolved** if that's the truth
- **Avoiding explanation** (the scene showed it; don't analyze it)

The audience should understand something they couldn't have understood from description.

---

## Outputs

The skill produces:

1. **Embodied Scene** - Multi-character dialogue with:
   - Distinct voice for each character
   - Internal thoughts/motivations revealed
   - Physical reality and power dynamics shown
   - Minimal narration (mostly dialogue and stage direction)
   - Collision of perspectives made visible

2. **Character Voice Key** (optional, if helpful) - Brief note on:
   - What distinguishes each character's speech
   - What each character wants/fears
   - Key assumptions driving their behavior

3. **Truth Revealed** - What the scene showed that description couldn't:
   - The gap between perspectives
   - The source of miscommunication or conflict
   - The human reality beneath the abstraction

---

## Error Handling

| Situation | Response |
|-----------|----------|
| Only one character in scene | Cannot apply skill; need minimum 2 participants |
| Characters sound identical | Revise - must have distinct voices, vocabulary, patterns |
| Scene reduces to good guy/bad guy | Revise - must show internal logic of all participants |
| Too much narration, not enough embodiment | Revise - replace description with character voice |
| No tension or collision | Scene may not benefit from this skill; tension is the material |
| Request to mock or dehumanize character | Refuse - skill requires showing humanity even in antagonists |

---

## Integration with Richard Pryor Expert

This skill operationalizes Pryor's character embodiment technique. When the Richard Pryor expert encounters described conflict or interaction, this skill provides the systematic process to transform it into lived, embodied scene.

**Pryor would use this skill to:**
- Take abstract "racial tension" and show the exact traffic stop interaction
- Transform "generational conflict" into specific argument between him and his grandmother
- Turn "addiction struggle" into the scene between addict-self and sober-self
- Make "police brutality" into the moment-by-moment collision of fears and misunderstandings

---

## Example Transformation

**INPUT:**
"There's often miscommunication between doctors and patients, especially when discussing serious diagnoses. Doctors are trained to be clinical, while patients are experiencing fear and uncertainty. This gap can lead to misunderstandings about treatment and prognosis."

**MULTI-CHARACTER SCENE OUTPUT:**

**Doctor:** "So, looking at your scan results, we've identified a mass. About 3 centimeters, right lobe. We'll need to do a biopsy to determine—"

**Patient:** (the word "mass" lands like a punch. Everything else is underwater noise. Mass. Tumor. Cancer. Dying.) "I'm sorry, what—how bad is it?"

**Doctor:** (sighs internally—they always jump ahead) "We don't know yet. That's what the biopsy will tell us. It could be benign, could be malignant. We won't know until—"

**Patient:** "But you think it's cancer."

**Doctor:** "I think it's a mass that requires further investigation. If we schedule the biopsy for next week—"

**Patient:** "Next week?" (Next week?? I could be dying right now and he's talking about scheduling?) "Can't we do it sooner?"

**Doctor:** (checks chart, already 20 minutes behind) "The imaging department is booked. Next Thursday is the earliest—"

**Patient:** "But what do I do until then? Do I—should I—" (voice cracking) "—what if it's spreading right now?"

**Doctor:** (speaking slowly, the way you'd explain to a child) "One week will not make a significant difference. The important thing is not to panic. Many masses are benign. Try not to worry until we have more information."

**Patient:** (Try not to worry. TRY NOT TO WORRY. He just said I have a mass and I should... not worry?) "Okay. Okay."

**Doctor:** (satisfied, standing up) "Any other questions? No? Great. Schedule with the front desk on your way out." (Next patient, next chart, next mass, same conversation)

**Patient:** (walking out, still hasn't heard anything after the word "mass") "Thank you, doctor."

---

**TRUTH REVEALED:**
The doctor is managing risk and following protocol—one week makes no clinical difference, and jumping to conclusions before the biopsy is medically premature. The patient heard "you might be dying" and everything after that was noise. The doctor is thinking in diagnostic workflows; the patient is experiencing existential terror. Neither is wrong. The gap between their realities creates the "miscommunication"—not because either failed, but because they're living the same moment from incompatible positions.

---

## Best Practices

1. **Each character must sound different** - If you can't tell who's talking without labels, revise
2. **Show don't tell emotional state** - Not "He was angry" but his voice, words, actions showing anger
3. **Internal logic must be valid** - Even antagonists have reasons that make sense to them
4. **Minimize narration** - The characters' voices do the work; don't interrupt with explanation
5. **Physical reality matters** - Who has power, who's vulnerable, what bodies are doing shapes the scene
6. **Let it be uncomfortable** - If there's no tension, there's no scene worth embodying

---

## Constraints

- **Minimum 2 characters required** - Can't build scene with only one perspective
- **Must show humanity in all participants** - Even "bad actors" must have internal logic
- **Cannot reduce to stereotypes** - Characters must be specific individuals, not stock types
- **Cannot resolve artificially** - If the tension is unresolved, don't force resolution
- **Narration must be minimal** - The scene lives in dialogue and embodiment, not description

---

## Success Criteria

Transformation is successful when:
- Each character has a distinct, recognizable voice
- The scene could be performed/acted without additional explanation
- Reader understands all perspectives even if they disagree with some
- The collision of perspectives creates the insight
- Description has been replaced by embodiment
- You can hear the voices, not just read the words
- The truth emerges from the scene itself, not from analysis of the scene

## Example

**Input:**
- input_data: [Specific example input]
- context: [Relevant background]

**Output:**

[Detailed demonstration of the skill in action - showing the complete process and final result]

**Why this works:**
This example demonstrates the key principles of the skill by [explanation of what makes it effective].