Beauty Karma

Beauty Karma

Overview

Beauty Karma turns an uploaded portrait photo into a playful camera-presence report. It uses MiniMax-M3 to analyze photo presentation factors such as lighting, composition, expression, clarity, atmosphere, and shareability. The skill returns a 1-10 lens score, a positive status label, short highlights, practical improvement tips, social-ready title/caption text, and a shareable report-card page.

Use cases:

• Generate a fun camera-presence report from a selfie, avatar, or portrait photo.
• Help choose or improve a profile photo.
• Create interactive social content and shareable mini-program cards.
• Provide lightweight, friendly, non-professional photo presentation feedback.

Quick Start

Create a JSON request file:

{
  "imagePath": "/absolute/path/to/photo.jpg",
  "userLabel": "Beauty Karma",
  "outputDir": "output/demo"
}

Run:

MINIMAX_API_KEY=your_key node scripts/run-skill.js request.json

Generated files:

• analysis.json: structured lens score and copy.
• yan-zhi-report.html: report-card page that can export a PNG.
• yan-zhi-report.svg: static fallback preview.

Inputs

• imagePath: local portrait photo path, absolute or relative to the skill folder.
• imageUrl: remote portrait photo URL.
• imageBase64: base64 image content, with or without a data:image/...;base64, prefix.
• imageNotes: optional visual notes used only if the configured MiniMax endpoint rejects image input.
• userLabel: optional footer label for the report card.
• outputDir: optional output directory.
• mock: optional; set to true for local demos without calling the model.

Model And Configuration

• Model: MiniMax-M3.
• Real analysis requires MINIMAX_API_KEY.
• Default endpoint: https://api.minimax.io/v1.
• Override with MINIMAX_BASE_URL if needed.

Safety Boundaries

The lens score is for entertainment and content creation only. It is not a measure of personal worth, scientific quality, or professional evaluation. Do not infer identity, age, ethnicity, health, wealth, or other sensitive attributes. Do not produce insults, degrading language, or appearance-value judgments. If the input is not a portrait or cannot be analyzed, ask for a clearer portrait photo.