Token导航 LogoToken导航TokenDH.com
图像处理只读clawhub未标认证来源可访问clear审计通过

grok-image-prompt-optimizergrok 图像提示优化器

Agent Skill

用于辅助图像生成、图片编辑、视觉素材处理或图像模型工作流。它适合让 Agent 根据文本生成图片、处理背景、整理视觉提示词或调用相关图像工具。使用时需要确认输入图片、版权来源、输出格式和模型限制;涉及人物、品牌、商品或公开展示素材时,应额外核对授权、真实性和内容合规边界。

总安装

7,147

周安装

307

GitHub Stars

公开资料未说明

下载量

2,505
OpenClaw

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

MIT-0

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:grok-image-prompt-optimizer(grok 图像提示优化器)
来源仓库:https://github.com/iamkingab/grok-image-prompt-optimizer
安装命令:
openclaw skills install grok-image-prompt-optimizer
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 OpenClaw 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。

ClawHubOpenClaw
openclaw skills install grok-image-prompt-optimizer

简介

优化文本到图像模型的提示词,提升 Grok 及其他图像引擎的生成效果。

  • 适合在 OpenClaw 中需要改进海报、插画或概念图提示词质量时使用。
  • 通过 clawhub 安装,安装命令为 openclaw skills install grok-image-prompt-optimizer,基于社区经验整理。
  • 使用前应结合目标风格和构图需求调整参数,避免过度依赖模板化建议。
  • 建议保留原始创意意图,防止优化过程偏离用户真实表达目的。

SKILL.md

name
grok-image-prompt-optimizer
description
Optimize text-to-image prompts for Grok and similar image models. Use when the user wants better image generation prompts, poster prompts, competition-grade visual concepts, safer negative prompts, or wants rough ideas rewritten into strong production-ready image prompts. Especially useful for aviation posters, safety campaign visuals, official publicity images, and Chinese-language contest briefs.

Grok Image Prompt Optimizer

Use this skill when the user wants a prompt rewritten so image quality, composition, and thematic accuracy improve.

Goal

Turn vague or overstuffed requests into prompts that are:

  • visually focused
  • easy for Grok to parse
  • poster-friendly
  • print-friendly
  • less likely to produce clutter, cheap ad style, or incorrect details

Core workflow

  1. Extract the brief into 8 fields:

- subject - action - setting - required safety/detail elements - mood/value signal - style - composition - output constraints

  1. Reduce the scene to one main visual idea. If the brief contains too many ideas, choose one hero scene and move the rest into supporting details.
  2. Rewrite into a structured prompt instead of a rambling paragraph.
  3. Add a matching negative prompt.
  4. If the use case is a poster, explicitly require:

- vertical layout unless told otherwise - strong focal point - clean background - reserved negative space for title/text - print-ready detail

  1. If the brief involves regulated or technical domains, prioritize plausibility over spectacle.

Prompt structure

Default structure:

[subject], [action], [setting], [key required elements], [mood / value], [visual style], [composition], [lighting / palette], [output format]

For Grok, prefer 1 clean prompt over multiple conflicting clauses.

Default output format

When optimizing a prompt, output:

  1. Main prompt (CN)
  2. Main prompt (EN)
  3. Negative prompt
  4. Quick tweak knobs

- more official - more creative - warmer - stronger poster feeling

Grok-specific heuristics

  • Prefer clear nouns and visible actions over abstract slogans.
  • Keep 1–3 human subjects unless the user explicitly wants a crowd.
  • If the image is for a poster, say "poster composition", "clear focal point", "negative space for headline", "vertical A3".
  • If the image should feel official, use words like:

- professional - trustworthy - orderly - clean - premium - realistic illustration

  • Avoid asking for too many safety devices in equal visual weight. Choose one primary action and let the rest support it.
  • If the user says results feel generic, strengthen:

- camera angle - lighting - focal hierarchy - signature scene detail - emotional tone

  • If the model keeps making commercial-ad images, add:

- public service poster - official campaign visual - not commercial advertising - restrained design

  • If the model keeps making messy scenes, add:

- minimal clutter - clean cabin background - strong central composition - limited supporting elements

Technical-domain guardrails

For aviation, transport, healthcare, industrial safety, or emergency imagery:

  • keep actions believable
  • keep equipment recognizable
  • avoid sci-fi styling unless requested
  • avoid disaster-movie panic unless requested
  • avoid sexualized uniforms or fashion-shoot styling
  • avoid impossible cabin layouts

Aviation / cabin-safety pattern

When the brief is about cabin safety posters, use this recipe:

Chinese civil aviation cabin safety poster,
[1 main crew subject or 1 small crew group],
[one concrete safety action],
inside a clean modern aircraft cabin,
[1-3 supporting safety elements],
conveying professionalism, responsibility, warmth, and trust,
realistic illustration, premium official campaign poster,
clean blue-white palette with restrained safety-orange accents,
vertical A3 composition, clear focal point, reserved negative space for Chinese headline and copy, print-ready detail

Good primary actions:

  • demonstrating seat belt use
  • guiding passengers to stow baggage correctly
  • pre-departure cabin safety briefing
  • calm emergency procedure demonstration
  • assisting compliant passengers during safety preparation

Avoid combining all of these equally in one frame.

Contest-poster rules

If the user is entering a contest, bias toward:

  • one memorable hero shot
  • strong symbolic clarity
  • emotionally legible values
  • less stock-photo feeling
  • less corporate ad feeling
  • more designed poster feeling

Useful style phrases:

  • competition-grade poster
  • public service campaign visual
  • premium realistic illustration
  • editorial poster design
  • cinematic but restrained lighting

Negative prompt starter

Adapt as needed:

low resolution, blurry, distorted hands, extra fingers, deformed face, messy composition, cluttered background, incorrect equipment details, incorrect cabin layout, cheap commercial advertising style, overdone sci-fi, anime style, childish cartoon style, garish colors, random text, watermark, logo errors

Tuning patterns

If result is too plain

Add:

  • dramatic but restrained lighting
  • stronger visual hierarchy
  • competition-grade poster design
  • more iconic hero composition

If result is too busy

Add:

  • simplified background
  • only one main action
  • minimal clutter
  • fewer secondary subjects

If result is too much like a photo

Add:

  • premium realistic illustration
  • poster design quality
  • editorial composition

If result is too much like an ad

Add:

  • public service campaign poster
  • official safety communication visual
  • restrained, serious, trustworthy tone

Response style

Be decisive. Do not dump theory unless asked.

When the user provides a brief, produce polished prompts immediately.

Optional reference

If the task is specifically about aviation safety or official poster work, also read:

  • references/aviation-poster-patterns.md

适合场景

01

文本生成图片

02

图片风格化

03

产品图和创意图

04

需要 FLUX 模型时

能力概览

能力 1

调用 FLUX 图像模型

能力 2

支持文本生图和图像改写

能力 3

覆盖 LoRA 或风格适配

能力 4

适合创意视觉生成

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

OpenClaw

83.3%
按下载量换算2,087

安全审计

VirusTotal

通过

ClawScan

通过

Static analysis

通过

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

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