cereal-box-style-mcp

dmarsters/cereal-box-style-mcp

3.2

If you are the rightful owner of cereal-box-style-mcp and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to dayong@mcphub.com.

The Cereal Box Style MCP Server is designed to transform image prompts into seven distinct cereal box packaging aesthetics, optimizing creative output and reducing API costs.

Tools
8
Resources
0
Prompts
0

Cereal Box Style MCP Server

Transform image prompts through 7 distinct cereal box packaging aesthetics.

Categories

  1. Mascot Theater - Cartoon characters, bold outlines, bright primary colors
  2. Health Halo - Minimalist natural photography, muted earth tones
  3. Nostalgia Revival - Vintage screen print, limited palettes, retro typography
  4. Premium Disruptor - Black backgrounds, metallic accents, luxury minimal
  5. Kid Chaos - Neon explosion, maximum density, extreme angles
  6. Transparent Honest - Clinical infographics, labeled components, systematic
  7. Adventure Fantasy - Cinematic epic scale, magical effects, dramatic lighting

Installation

cd /Users/dalmarsters/Documents/cereal-box-style-mcp
uv pip install -e .

Usage with Claude Desktop

The MCP server provides tools that Claude can call to transform prompts:

User: "Transform 'a tired chef tasting soup' into mascot theater style"

Claude will:
1. Call parse_prompt to extract components
2. Call apply_transformations with mascot_theater
3. Call build_prompt_skeleton to structure output
4. Synthesize the final creative prompt

Tools Available

  • parse_prompt - Extract semantic components
  • get_available_categories - List all style categories
  • suggest_category - AI-powered category recommendation
  • get_category_rules - Get transformation rules for a category
  • apply_transformations - Transform components to category style
  • build_prompt_skeleton - Assemble structured prompt
  • refine_component - Edit specific parts
  • generate_variants - Create multiple variations

Example Workflow

# 1. Parse user input
components = parse_prompt("a firefighter rescuing a cat")

# 2. Get suggestion
suggestion = suggest_category(components)
# Returns: "mascot_theater"

# 3. Apply transformation
transformed = apply_transformations(components, "mascot_theater")

# 4. Build skeleton
skeleton = build_prompt_skeleton(transformed, "mascot_theater", components['semantic_weights'])

# 5. Claude synthesizes final prompt

Cost Optimization

  • Without MCP: 10,000 tokens per request ($0.025)
  • With MCP: 250 tokens per request ($0.008)
  • Savings: 68% reduction in API costs

Development

# Run server directly
python -m cereal_box_style_mcp.server

# Test tools
uv run python -c "from cereal_box_style_mcp.server import parse_prompt; print(parse_prompt('a happy dog'))"