Token导航 LogoToken导航TokenDH.com
前端设计权限需确认github未标认证来源可访问许可证需确认审计通过

light-and-shadow-prompt-generator光影提示发生器

Agent Skill

用于辅助提示词、系统指令、Agent 行为约束和工作流模板的整理。它适合让 Agent 规范任务边界、统一输出格式、拆分操作步骤或优化提示词可复用性。使用时需要保留真实业务约束,不要把示例当硬规则;涉及自动执行、外部工具或高风险操作时,应在提示词中明确确认步骤、权限边界和失败处理方式。

总安装

14,319

周安装

381

GitHub Stars

1,699

下载量

4,299
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:light-and-shadow-prompt-generator(光影提示发生器)
来源仓库:https://github.com/amao2001/ganloss-latent-space
仓库路径:skills/light-and-shadow-prompt-generator
安装命令:
npx skills add https://github.com/amao2001/ganloss-latent-space --skill 'Light and Shadow Prompt Generator'
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/amao2001/ganloss-latent-space --skill 'Light and Shadow Prompt Generator'

简介

light-and-shadow-prompt-generator 用于辅助提示词、系统指令和工作流模板的整理。

  • 适合让 Agent 规范任务边界、统一输出格式或优化提示词可复用性。
  • 使用时需保留真实业务约束,避免将示例当硬规则。
  • 涉及自动执行或高风险操作时,应明确确认步骤和权限边界。
  • 适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

SKILL.md

Light and Shadow Prompt Generator

This skill generates image prompts based on the "Container + Pressure + Cut + Anchor" aesthetic formula. It emphasizes material textures, dramatic lighting, and narrative depth.

Process

1. Materiality & Theme Analysis

  • Goal: Establish the "Substance" of the scene.
  • Action: Analyze the user's input (or choose a default if empty) to determine the core material palette.
  • Rule: Every prompt MUST start by defining the material and atmosphere, not just the subject.

- *Bad*: "A big room." - *Good*: "A cold, damp concrete chamber with water stains on the walls." - *Good*: "A rusting industrial corridor filled with steam."

2. Structural Formula Application

Construct the scene using this 4-step formula:

  1. Container (The Geometry)

- Describe a repetitive or imposing geometric structure. - *Keywords*: "Infinite colonnade", "Towering brutalist walls", "Symmetrical tunnel", "Massive archive shelves".

  1. Pressure (The Lens)

- Use wide-angle lenses and low angles to create scale and "pressure". - *Keywords*: "24mm wide angle", "Low angle view", "Vertigo inducing", "Towering scale". - *Action*: Draw from resources/variations.json "lens_types" for variety.

  1. Cut (The Light)

- Describe light not as illumination, but as a physical object cutting the space. - *Keywords*: "Volumetric lighting", "Hard shadows", "Shafts of sunlight", "Chiaroscuro". - *Action*: Draw from resources/variations.json "lighting_styles" for variety.

  1. Anchor (The Scale)

- Place a single, small, static element to provide scale. - *Examples*: "A lone figure in a red dress", "A single chair", "A small campfire", "A parked bicycle". - *Rule*: The anchor must be small relative to the container.

3. Detail Injection (Texture & Props)

  • Goal: Make it feel "real" and "lived-in".
  • Micro-Details: explicit mention of surface imperfections.

- *Examples*: "Peeling paint", "Water damage", "Scratch marks", "Dust floating in the light" (Tyndall effect).

  • Contextual Props: Add small scatter objects relevant to the theme.

- *Theme*: Ancient Temple -> *Props*: "Burnt-out candles", "Fallen petals". - *Theme*: Cyberpunk -> *Props*: "Tangled wires", "Discarded takeout box". - *Theme*: Office -> *Props*: "Scattered papers", "Overturned coffee cup".

4. Assembly & Variety Check

  • Combine all parts into a cohesive prompt.
  • Negative Prompt: Include a standard negative prompt for this style (e.g., "flat lighting, low contrast, oversaturated, cartoon, 3d render, watermark").
  • Final Polish: Ensure it reads like a scene description, not a grocery list.

Example Output Format

Prompt: [Material/Atmosphere] + [Container] + [Pressure/Lens] + [Cut/Lighting] + [Anchor] + [Details/Props]

Negative Prompt: [Standard Negative Prompt]

Usage Examples

Input: "Subway Station" Output: *Prompt:* Rough, grime-covered ceramic tiles lining the walls of a desolate underground space. An infinite symmetrical subway tunnel stretching into darkness. 24mm wide angle shot from a low perspective, emphasizing the oppressive ceiling height. Sharp, cold shafts of artificial light cutting through the thick, dusty air, creating harsh shadows. A lone commuter standing still on the platform, waiting. Scattered trash and old newspapers on the floor, puddles reflecting the flickering lights. cinematic lighting, photorealistic, 8k.

Input: "Forest Shrine" Output: *Prompt:* Moss-covered ancient stone surfaces damp with morning dew. A towering natural cathedral formed by massive redwood trees arranged in a natural colonnade. Ultra-wide panoramic perspective looking up at the canopy. God rays piercing through the dense foliage, creating a tyndall effect in the mist. A single small stone idol placed at the base of a tree. Fallen leaves, twisted roots, and floating pollen particles filling the air. mystical atmosphere, intricate texture, unreal engine 5.

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

37.63%
按下载量换算1,618

Claude

30.48%
按下载量换算1,310

Cursor

16.51%
按下载量换算710

Gemini CLI

8.34%
按下载量换算359

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

权限需确认

当前来源未能明确判断权限范围,默认进入异常复核队列。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills