the-clawcap（爪帽）

🦞🧠 The ClawCap — Avatar Accessory Skill

When the user wants to add a hat, headwear, or accessory to an avatar image, use this skill.

What It Does

ClawCap uses a 3-stage AI pipeline to add accessories to any avatar:

1. VLM Fingerprint — Gemini Vision analyzes the avatar's art style, lighting, color palette, and head position
2. Mask Generation — Automatically creates a pixel-precise placement mask using Pillow/NumPy
3. AI Inpainting — Gemini generates the accessory that perfectly matches the original style

Supported Styles

• 📷 Real photos
• 🎨 Anime / 2D illustrations
• 🧊 3D renders
• 👾 Pixel art / NFT avatars

How to Use

As MCP Skill (Claude Desktop)

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "clawcap": {
      "command": "python",
      "args": ["mcp_server.py"],
      "cwd": "/path/to/The-ClawCap"
    }
  }
}

Then tell Claude: *"给这个头像戴上一顶龙虾帽"* with an image attached.

As Web API

POST http://localhost:8000/api/skill/alpha-equip
Content-Type: application/json

{
  "image_base64": "<base64 encoded image>",
  "accessory_prompt": "a red lobster claw beanie hat"
}

Live Demo

🌐 Try it online (Rate limited: 3 requests/minute per IP)

Requirements

• Python 3.10+
• GEMINI_API_KEY environment variable (get one at https://aistudio.google.com/apikey)
• Install deps: pip install -r requirements.txt

Repository

🔗 GitHub: The-ClawCap