Review Story

name: review-story description: > Multi-provider creative writing critique for stories, screenplays, and narratives. Analyzes structural, emotional, craft, and persona dimensions. Integrates with Horus's Theory of Mind for voice consistency validation. allowed-tools: [Bash, Read, Write, Task] triggers:

• review story
• critique story
• story review
• review screenplay
• critique screenplay
• review narrative
• review my story
• review draft
• story feedback
• writing critique metadata: short-description: Multi-provider creative writing critique with ToM alignment author: "Horus" version: "0.1.0"

provides: - review-story composes: [, task-monitor]

review-story

Multi-provider creative writing critique for Horus persona. Analyzes stories across four dimensions and returns structured feedback for iterative refinement.

Critique Dimensions

Supported Providers

Quick Start

cd .pi/skills/review-story

# Single provider review
./run.sh review draft.md --provider claude --emotion rage

# Multi-provider review (recommended for final drafts)
./run.sh review draft.md --providers claude,codex --emotion sorrow

# Focus on specific dimensions
./run.sh review draft.md --provider claude --focus structural,emotional

# Review with Horus persona validation
./run.sh review draft.md --provider claude --validate-persona

Commands

review - Critique a Story

./run.sh review <story_file> [OPTIONS]

Options:
  --provider TEXT       Single provider (claude, codex, gemini, copilot)
  --providers TEXT      Comma-separated list for multi-provider review
  --emotion TEXT        Intended emotion (rage, sorrow, camaraderie, regret)
  --focus TEXT          Dimensions to focus on (structural, emotional, craft, persona)
  --validate-persona    Validate against Horus voice patterns
  --output-dir PATH     Output directory for critique files
  --format TEXT         Output format (json, markdown)

compare - Compare Multiple Drafts

./run.sh compare draft_v1.md draft_v2.md --dimension emotional

synthesize - Combine Multi-Provider Critiques

./run.sh synthesize critique_claude.json critique_codex.json --output synthesis.md

Output Format

{
  "provider": "claude",
  "model": "claude-sonnet-4-20250514",
  "story_file": "draft_v1.md",
  "intended_emotion": "rage",

  "structural": {
    "score": 7,
    "issues": [
      {"location": "middle", "issue": "Pacing drags in act 2", "severity": "medium"},
      {"location": "climax", "issue": "Needs more foreshadowing", "severity": "high"}
    ],
    "strengths": ["Strong opening hook", "Clear character motivation"],
    "suggestions": ["Add tension beat at paragraph 15", "Foreshadow betrayal earlier"]
  },

  "emotional": {
    "intended": "rage",
    "achieved": "frustration",
    "alignment_score": 0.6,
    "tom_pattern": "DDL_intensity",
    "issues": ["Build too gradual - DDL rage is explosive after slow burn"],
    "suggestions": ["Amplify with longer setup, then sudden release"]
  },

  "craft": {
    "prose_score": 8,
    "dialogue_score": 6,
    "sensory_score": 7,
    "issues": ["Dialogue feels too modern for Warmaster voice"],
    "suggestions": ["Add more tactical metaphors", "Replace contractions"]
  },

  "persona": {
    "horus_voice_score": 0.7,
    "issues": ["Too much explanation, not enough contempt", "Missing resentment undertones"],
    "tactical_mask_detected": "Tywin",
    "suggestions": ["Add bitter aside about imprisonment", "More sardonic observations"]
  },

  "overall": {
    "score": 7.2,
    "ready_for_next_draft": true,
    "priority_fixes": ["Emotional buildup", "Persona voice"]
  },

  "taxonomy": {
    "bridge_tags": ["Corruption", "Loyalty"],
    "collection_tags": {
      "function": "Confrontation",
      "domain": "Primarch",
      "thematic_weight": "Tragedy"
    },
    "confidence": 0.75,
    "worth_remembering": true
  }
}

Federated Taxonomy Integration

All review-story outputs include taxonomy metadata for multi-hop graph traversal:

Bridge Attributes (Cross-Collection Connectors)

Multi-Hop Example

[Story Critique]                      [Code Lesson]
bridge_tags: ["Resilience"]     →    bridge_tags: ["Resilience"]
"Dorn's endurance at Terra"          "Fault-tolerant retry logic"
        ↘                           ↙
          [Query: "Endurance patterns"]
                    ↓
         Both retrieved via shared
         "Resilience" bridge attribute

Integration with create-story

The create-story skill calls review-story between drafts:

DRAFT 1
    ↓
/review-story draft_1.md --provider claude --emotion rage
    ↓
[Synthesize critique into actionable notes]
    ↓
DRAFT 2 (addresses structural + emotional issues)
    ↓
/review-story draft_2.md --providers claude,codex --focus craft,persona
    ↓
FINAL DRAFT

Horus Voice Patterns (Persona Validation)

When --validate-persona is enabled, the critique checks against:

Tactical Masks (from HORUS_PERSONA.md)

Emotional Patterns (from HORUS_TOM_SYSTEM.md)

Example Session

$ ./run.sh review draft_v1.md --provider claude --emotion rage --validate-persona

[REVIEW] Analyzing draft_v1.md...
[REVIEW] Provider: claude (claude-sonnet-4-20250514)
[REVIEW] Intended emotion: rage

[STRUCTURAL] Score: 7/10
  ⚠ Middle section pacing drags (lines 45-78)
  ⚠ Climax needs stronger foreshadowing
  ✓ Strong opening hook
  ✓ Clear character motivation

[EMOTIONAL] Alignment: 60%
  Intended: rage → Achieved: frustration
  Pattern: DDL_intensity not fully realized
  → Suggestion: Longer slow burn before explosive release

[CRAFT] Score: 7.5/10
  Prose: 8 | Dialogue: 6 | Sensory: 7
  ⚠ Dialogue too modern for Warmaster
  → Replace contractions, add tactical metaphors

[PERSONA] Horus Voice: 70%
  Detected mask: Tywin (Authority)
  ⚠ Missing resentment undertones
  ⚠ Too explanatory, not enough contempt
  → Add bitter aside about imprisonment

[OVERALL] 7.2/10 - Ready for Draft 2
  Priority fixes: Emotional buildup, Persona voice

Output: review_output/claude_draft_v1.json

Environment

Dependencies

• Python 3.11+
• Access to at least one provider (claude, codex, gemini, or copilot)
• Optional: Access to Horus persona files for voice validation

Memory Integration

The memory_integration.py module provides automatic memory recall and learning:

Pre-hook: recall_prior_critiques(title_or_genre, k=5)

• Recalls past story critiques from memory before starting a new review
• Surfaces recurring issues, patterns, and what improved between drafts
• Uses taxonomy bridge tags for cross-collection recall

Post-hook: learn_critique(title, genre, provider, findings, strengths, weaknesses, score)

• Learns critique findings to memory after review completion
• Stores: critique summary, individual strengths, individual weaknesses, full findings
• Tags: ["story_review", genre] + bridge_tags
• Bridge keywords: Precision (craft, prose, pacing), Resilience (arc, resolution, character growth), Fragility (plot hole, pacing issue, weak dialogue), Stealth (subtext, theme, symbolism)

Graceful Degradation

• All memory operations use try/except ImportError
• If common.memory_client is unavailable, hooks are silently skipped
• Taxonomy loaded dynamically via importlib.util to avoid name conflicts

Related Skills