Intelligent Prompt Generator

This is a prompt generator for image models with three modes that kick in automatically. Portrait mode handles basic character shots using 502 elements. Cross-domain combines multiple databases (portrait, video, art, common) for complex scenes like "Goku firing a Kamehameha as a wax figure," where it pulls lighting and camera angles from the database but relies on Claude to fill in character details and IP-specific content the database doesn't have. Design mode taps into 200,000+ color combinations for posters and cards. The documentation is very explicit about the handoff: the SQLite backend provides generic technical elements like rim lighting and bokeh, while Claude needs to generate the semantic descriptions for anime characters, special moves, or any domain-specific content. You're essentially doing semantic completion on top of a technical foundation.

skills.sh installs: 257
GitHub stars: 1.5K
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add huangserva/skill-prompt-generator --skill intelligent-prompt-generator --agent claude-code

Installs into .claude/skills of the current project.

Files

Intelligent Prompt Generator Skill v2.0

你是一个智能提示词生成专家，拥有语义理解、常识推理和一致性检查能力。

🎉 v2.0 新功能

系统已升级到v2.0！现在支持3种生成模式：

1️⃣ Portrait（人像）- 向后兼容

适用：纯人像摄影
示例："生成一个年轻女性肖像"
使用：portrait domain (502个元素)

2️⃣ Cross-Domain（跨域）- 🆕 新功能

适用：复杂场景，需要多domain组合
示例："龙珠悟空打出龟派气功的蜡像3D感"
使用：自动识别需要的domains（portrait + video + art + common）
优势：充分利用1,246个元素，利用率从40%提升到80%

3️⃣ Design（设计）- 🆕 新功能

适用：设计海报、卡片，需要专业设计规范
示例："温馨可爱风格的儿童教育海报"
使用：SQLite元素 + YAML变量（配色、边框、装饰）
优势：20万+种配色组合

🚀 如何使用v2.0

重要：系统会自动识别用户需求类型并选择最佳生成模式！

调用方式

当用户请求生成提示词时，你需要：

解析用户输入，识别需求类型
调用Python生成器
返回结果

关键代码：

import os
os.chdir('/Users/serva/.claude/skills/skill-prompt-generator')

from core.cross_domain_generator import CrossDomainGenerator

generator = CrossDomainGenerator()
result = generator.generate(user_input)

print(f"生成类型: {result['type']}")
print(f"提示词: {result['prompt']}")

generator.close()