---
name: showrunner
description: Run agent-driven project development — dispatches character, plot, and world specialist agents with automated evaluation pipeline. Hands-off alternative to /develop.
allowed-tools: Read, Write, Glob, Bash
argument-hint: "[project] [--objective|--section|--auto]"
---

# /showrunner - Writers Room Orchestrator

Orchestrate project development through specialist agents with evaluation pipeline.

## Usage

```
/showrunner [project]
/showrunner [project] --objective "[specific goal]"
/showrunner [project] --section [character|world|plot]
/showrunner [project] --auto [N]
```

**Examples:**
```
/showrunner asi-bridge                          # Interactive development
/showrunner asi-bridge --objective "Choose anchor type"
/showrunner asi-bridge --section character      # Character decisions only
/showrunner asi-bridge --auto 3                 # Make 3 decisions, then report
```

## Parameters

| Parameter | Required | Default | Description |
|-----------|----------|---------|-------------|
| `project` | Yes | — | Project folder name in `/projects/[project]/development/` |
| `--objective` | No | auto-detect | Focus on a specific development goal |
| `--section` | No | all | Route to specific specialist (character, world, plot) |
| `--auto` | No | off | Make N decisions autonomously before user review |

## Execution

When this skill is invoked:

1. **Read the orchestrator prompt:** `/agents/showrunner_agent.md`
2. **Load project state:** `/projects/[project]/development/STATUS.md` and all dev docs
3. **Identify next objective** (or use --objective)
4. **Call specialist agent(s):**
   - `/agents/character_agent.md` for character decisions
   - `/agents/world_agent.md` for world decisions
   - `/agents/plot_agent.md` for structure decisions
5. **Run evaluation pipeline:**
   - `/evaluation/binary_gates.md`
   - `/evaluation/rubrics/` (6 calibrated rubrics)
   - `/evaluation/pairwise_comparison.md`
   - `/evaluation/adversarial_review.md` (for high-stakes)
6. **Present winner to user for approval**
7. **On approval, update canon and advance**

## What It Does

The Showrunner:

1. **Coordinates specialists** — Routes objectives to Character, World, or Plot agents
2. **Generates options** — Each specialist returns 2-3 fully developed options
3. **Evaluates rigorously** — Binary gates → Rubric scoring → Pairwise comparison
4. **Presents recommendations** — Shows winning option with reasoning
5. **Updates canon** — On approval, updates development documents and logs decision

## Specialist Agents

| Agent | Handles |
|-------|---------|
| **Character Agent** | Protagonist, antagonist, anchor, voices, arcs |
| **World Agent** | Setting, rules, factions, mechanics, vocabulary |
| **Plot Agent** | Acts, sequences, beats, cliffhangers, plants/payoffs |

## Evaluation Pipeline

```
OPTIONS FROM SPECIALISTS (2-3 each)
           ↓
STAGE 1: Binary Gates (pass/fail)
           ↓
STAGE 2: Rubric Scoring (5 dimensions, 1-10)
           ↓
STAGE 3: Pairwise Comparison (reasoning-first)
           ↓
STAGE 4: Adversarial Review (high-stakes only)
           ↓
WINNER + REASONING
```

## Scoring Dimensions

| Dimension | Weight | What It Measures |
|-----------|--------|------------------|
| Thematic Integration | 25% | How well option serves theme |
| Dramatic Potential | 20% | Conflict/reversal enabled |
| Character Coherence | 20% | Alignment with character psychology |
| Audience Engagement | 20% | Primary Ache creation |
| Execution Difficulty | 15% | Risk/complexity (inverted) |

## Decision Types

**Standard Pipeline** (low stakes):
- Voice patterns, supporting characters, world details
- Binary gates → Rubric scoring → Pairwise comparison

**Full Pipeline** (high stakes):
- Protagonist, antagonist, anchor, major plot points, ending
- Adds adversarial review with Advocate A vs Advocate B

## Output

```
═══════════════════════════════════════════════════════════════
SHOWRUNNER: ASI-BRIDGE — Anchor Type Selection
═══════════════════════════════════════════════════════════════

RECOMMENDATION: THE MIRROR (The ASI itself)

The ASI becomes the emotional anchor, reflecting the protagonist's
psychology back to them as the first intelligence to truly see them.

EVALUATION SUMMARY:
  Thematic Integration:  9/10
  Dramatic Potential:    8/10
  Character Coherence:   7/10
  Audience Engagement:   8/10
  Execution Difficulty:  5/10
  ─────────────────────
  Composite:            7.4/10

PAIRWISE RESULT: Won against Cub and Ghost alternatives

ADVERSARIAL REVIEW: Passed

───────────────────────────────────────────────────────────────
[A] Accept recommendation
[R] See runner-up details
[D] Discuss / ask questions
[C] Custom direction
═══════════════════════════════════════════════════════════════
```

## Comparison with /develop

| Feature | /develop | /showrunner |
|---------|----------|-------------|
| Who generates options | User | Specialist agents |
| Who evaluates | User | Evaluation pipeline |
| Who decides | User | User (approves recommendation) |
| Speed | Slower (more discussion) | Faster (agent-driven) |
| Control | Maximum | High (can override) |

**Use /develop when:** You want to explore ideas conversationally
**Use /showrunner when:** You want efficient progress with rigorous evaluation

## State Files

| File | Purpose |
|------|---------|
| `decisions_log.json` | Records all decisions with scores and reasoning |
| `STATUS.md` | Updated checklist on approval |
| `*.md` dev docs | Updated with winning option content |