---

name: "@tank/character-creator" description: | Create complete characters — AI-generated visual sheets (turnarounds, expressions, poses) plus written character bibles. Discovery interview, archetype-to-visual mapping, AI generation via Flux Kontext and GPT-Image-1. Any style: anime, realistic, pixel art, cartoon, vector. Triggers: character design, character sheet, create a character, expression sheet, character bible, OC design, game character, NPC design, villain design, pose sheet, model sheet.

Character Creator

Core Philosophy

1. Personality first, aesthetics second. Every visual decision — shape, color, silhouette, costume — flows from the character's personality and archetype. Design without personality produces forgettable characters.
2. Generate, don't describe. Use AI image generation to produce actual character art — turnarounds, expressions, poses. Ship reference sheets, not text descriptions of what a character could look like.
3. Separate views, composite later. Generate each pose/angle at full resolution independently. Multi-view single-image generations lose face detail and consistency. Assemble the final sheet in post.
4. Consistency through reference chaining. Generate the canonical front view first, then use it as the reference image for every subsequent generation. The front view is the source of truth.
5. Complete package or nothing. A character is not done until it has both visual sheets (turnaround + expressions minimum) and a written spec (character bible). One without the other is incomplete.

Quick-Start: Create a Character

"I have a character idea"

1. Discover — Run the discovery interview (5-10 questions) to extract personality, motivation, role, world, and visual direction. -> See references/character-discovery.md
2. Define — Map personality to archetype, then archetype to visual DNA: shape language, color palette, silhouette, proportions. -> See references/archetype-visual-system.md -> See references/shape-color-theory.md
3. Design — Specify the visual deliverables: which sheet types, how many views, expression count, pose selection. -> See references/character-sheet-components.md
4. Generate — Create the character art using AI image generation: canonical image → turnarounds → expressions → poses. -> See references/ai-generation-pipeline.md -> See references/prompt-templates.md
5. Deliver — Write the character bible and assemble the final package: visual sheets + written spec + structured data. -> See references/character-bible-output.md

"I need a character sheet for an existing character"

1. Skip discovery — gather the character's existing description, personality, and visual references.
2. Write the Character DNA block (50-80 word visual description). -> See references/prompt-templates.md for DNA block template
3. Generate sheets directly using the AI pipeline. -> See references/ai-generation-pipeline.md

"I need a quick NPC or minor character"

1. Run the abbreviated interview (Q1, Q2, Q3, Q4, Q9 only).
2. Use the Quick Character Card format instead of full bible. -> See references/character-bible-output.md for quick card template

Decision Trees

Art Style Selection

User Signal	Recommended Style	Prompt Approach
Game characters, stylized	Cartoon / Stylized	Bold outlines, exaggerated features, flat colors
Anime or manga project	Anime	Cel shading, large eyes, clean linework
AAA game or film concept	Realistic	Detailed rendering, anatomical proportions
Mobile or retro game	Pixel Art	Limited palette, clean pixels, retro aesthetic
Web/app mascot, icons	Vector / Flat	Solid colors, no gradients, SVG-ready
Concept art, illustration	Painterly	Visible brushstrokes, rich values

Sheet Type Selection

Need	Minimum Deliverable	Full Deliverable
Game character	Turnaround (3 views) + T-pose	5-view turnaround + expressions + poses + props
Animation character	5-view turnaround + expressions	+ lip sync chart + hair physics notes
Story/novel character	Expression sheet + character bible	+ turnaround + costume variants
NPC / minor character	Front view + quick card	+ idle pose + 2 expressions
Mascot / brand character	Turnaround + expression sheet	+ pose library + style guide

API Selection

Task	Recommended API	Cost	Why
Canonical front view	GPT-Image-1 or Flux	$0.04-0.10	Best initial generation quality
Turnaround views	Flux Kontext Pro (fal.ai)	$0.04/view	Best character consistency from reference
Expression sheet	GPT-Image-1	$0.04-0.10	Conversational context preserves identity
Pose variations	Flux Kontext Pro (fal.ai)	$0.04/pose	Reference chaining maintains consistency
Vector/SVG output	Recraft V4 (fal.ai)	$0.08	Native SVG generation
Face identity lock	Flux-PuLID (fal.ai)	$0.005	When face drifts between views

Scope Selection

Character Importance	Interview Depth	Visual Package	Written Spec
Protagonist / Antagonist	Full (10 questions)	Full sheet set	Full character bible
Supporting character	Standard (7 questions)	Turnaround + expressions	Standard bible
NPC / minor character	Quick (5 questions)	Front view + 2 expressions	Quick character card
Background / extra	Skip interview	Front view only	One-line description

Authentication

fal.ai (Flux Kontext, Recraft, Flux-PuLID):

export FAL_KEY="YOUR_FAL_KEY"
npm install @fal-ai/client

OpenAI (GPT-Image-1):

export OPENAI_API_KEY="YOUR_KEY"
npm install openai

Reference Files

File	Contents
references/character-discovery.md	Discovery interview (10 questions), personality profiling (5 dimensions), archetype quick-select, character essence statement formula, red flags, quality gates
references/archetype-visual-system.md	All 12 Jungian archetypes with complete visual DNA: shape language, color palettes (hex), silhouette, proportions, costume, movement, AI prompt hints, blending rules, cast-level design
references/shape-color-theory.md	Shape language (circle/square/triangle), combining shapes, silhouette design, color psychology per color, 60-30-10 palette rule, palette construction from archetype, professional color tests, color scripting for arcs
references/character-sheet-components.md	Turnaround specs (3-5 views), expression sheet (6-12 emotions), proportion guide, pose sheet, detail callouts, props, color palette documentation, annotations, industry format differences (games/animation/comics), complete deliverable checklist
references/ai-generation-pipeline.md	5-step generation pipeline, API selection and auth setup, canonical image generation, turnaround/expression/pose generation with Flux Kontext and GPT-Image-1 code examples, consistency techniques (ranked), assembly, cost estimation, troubleshooting
references/prompt-templates.md	Character DNA block template, turnaround/expression/pose prompt templates, style-specific variants (anime/realistic/cartoon/pixel/vector/painterly), negative constraints, iteration diagnosis, complete worked example
references/character-bible-output.md	10-section character bible template, JSON schema, Markdown spec template, quick character card format, output file organization, medium-specific adaptations (games/animation/comics/TTRPG), quality checklist