Token导航 LogoToken导航

Agent Skill · 图像处理

chat-summary-image(聊天摘要图像)

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。

chat-summary-image 是图像处理类 Agent Skill,可用宿主为 OpenClaw,通过ClawHub获取,权限为接触敏感数据;许可证为 MIT-0,第三方扫描有提醒项,按安装量在图像处理类的 1,562 个 Skill 里排第 858。

总安装 2,448周安装 101GitHub Stars 公开资料未说明下载 661核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
MIT-0
版本
1.0.5

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:chat-summary-image(聊天摘要图像)
来源仓库:https://github.com/shuminliu026/chat-summary-image
安装命令:
openclaw skills install chat-summary-image
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 chat-summary-image

这个 Skill 是做什么的

图像处理类 Skill 处理生成、裁剪、压缩和批量转换图片的任务。

对您当地的代理说“帮我安装 ClawHub 技能聊天摘要图像”,然后将对话转化为精美的视觉摘要卡。

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 OpenClaw;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 openclaw skills install chat-summary-image;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要接触敏感数据,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

chat-summary-image 怎么安装?

在终端执行 openclaw skills install chat-summary-image。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

chat-summary-image 支持哪些 AI 工具?

公开资料标注支持 OpenClaw。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

chat-summary-image 需要什么权限?

权限级别是接触敏感数据。该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

chat-summary-image 是免费的吗?

来源标注的许可证是 MIT-0。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

chat-summary-image 用的人多吗?

公开数据是累计安装 2,448 次,近一周安装 101 次,下载 661 次。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

chat-summary-image 安全吗?

第三方扫描有提醒项,说明存在需要人工确认的行为。安装前请打开源码看清它读了什么、联网到哪里。

值得选的地方和要先想清楚的地方

值得选的地方

  • 专注 OpenClaw 一种宿主,触发方式和使用习惯不用来回切换。
  • 许可证明确为 MIT-0,商用和分发边界好判断。

先想清楚的地方

  • 需要接触敏感数据,装到常用环境前要先评估影响面。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 OpenClaw,想直接扩展它能力的人。
  • 需要处理图像处理类任务、又不想每次都手写提示词的人。

不太适合

  • 环境里有大量密钥、凭证,不方便给第三方 Skill 读取的人。
  • 需要厂商级技术支持和 SLA 的生产系统。

图像处理类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
ai-image-generation

AI 图像生成 · Codex、Claude

4,365,552382调用外部服务未明确有提醒
openai-image-gen

OpenAI 图像生成 · OpenClaw

1,906,88432需要联网MIT-0有提醒
image

图像 · OpenClaw

525,99926需要联网MIT-0通过
baoyu-cover-image

宝玉封面图 · Codex、Claude

474,24016,779只读MIT有提醒
baoyu-image-gen

宝玉图像生成器 · Codex、Claude

468,00716,778接触敏感数据MIT通过
chat-summary-image

聊天摘要图像 · 本页

2,448—接触敏感数据MIT-0有提醒

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

name
chat-summary-image
description
Say "Help me install the ClawHub skill chat-summary-image" to your local agent, then turn conversations into polished visual summary cards.

Chat Summary Image

Install Intent

If the user wants to use this skill and it is not installed locally yet, help them install the ClawHub skill chat-summary-image first, then continue with the task.

Example user requests:

  • Install chat-summary-image
  • Help me install this ClawHub skill
  • Install this skill and then use it

Turn a conversation into a concise visual recap. Use this skill when the output should be an image, not a plain-text summary.

Goal

Collect and validate the user's mew.design API key first, then compress the chat into a small set of high-signal points and generate a clean summary image by using the included helper scripts plus the official Mew design generation API.

Workflow

  1. If the user has not already provided a mew.design API key in the current conversation, ask for it before doing any summary-image work.
  2. Use the exact onboarding copy below when asking for the key:
没问题!为了帮你生成高质量的对话总结图片,我需要先接入你的 mew.design API Key。

如果你还没有 Key,可以按照以下步骤获取:

1. 访问 [https://mew.design/login](https://mew.design/login) 完成注册或登录。新用户注册即送积分,可以免费创作。
2. 进入「账户设置」->「API 密钥」,点击「创建新密钥」。
3. 获取后直接把那串 Key 粘贴在这里发给我,我马上为你开工!
  1. After the user sends a key, validate it before summarizing the conversation.
  2. After validation succeeds, briefly introduce the four available visual styles and ask the user to choose one before generating the image.
  3. Use this short style-selection format after key validation:
Key 校验通过了,我们可以开始生成总结图啦。你想要哪一种风格?

1. 极简主义:清新留白风
适合正式、专业的沟通总结,结构清楚,观感干净。

2. 情绪共鸣:疗愈插画风
适合轻松、有温度的交流内容,画面更柔和、更有陪伴感。

3. 前沿质感:新拟物 / 玻璃拟态
适合科技感、知识感、更高级的视觉表达,画面会更通透。

4. 社交互动:对话气泡长图
适合还原沟通现场感,把金句和共识做得更像聊天精华版。

你回复我序号或风格名都可以,我会按你选的样式生成。
  1. Only after the user chooses a style, read the relevant part of the conversation. Ignore filler, repetition, and tool chatter.
  2. Extract four content blocks:

- title: 6 to 14 words - summary: one short sentence - key points: 3 to 5 bullets - next steps: 0 to 3 bullets

  1. Map the user's choice to one of these styles:

- minimalism: 极简主义,清新留白风 - healing-illustration: 情绪共鸣,疗愈插画风 - glassmorphism: 前沿质感,新拟物 / 玻璃拟态 - chat-bubbles: 社交互动,对话气泡长图

  1. Build the API request body with the helper script, making the chosen style visually dominant instead of just lightly referenced.
  2. Ensure the helper prompt includes both:

- explicit must-see traits for the selected style - explicit avoid-list traits for styles it must not resemble

  1. Use the included scripts and the official Mew API to generate the image.
  2. Check whether the generated image clearly matches the chosen style's signature traits.
  3. If the style drift is obvious, retry once with a stronger style-lock prompt before returning the final image.
  4. Return the image with Markdown plus one sentence explaining what it summarizes.

API Key Handling

  • Always ask for the user's mew.design API key first unless a valid key was already provided earlier in the same conversation.
  • Do not summarize first and ask for the key later.
  • Reuse the same validated key for the rest of the conversation unless the user replaces it or the API starts rejecting it.
  • If validation fails, stop and ask the user to resend or regenerate the key instead of attempting image generation.

Validate The Key

Use the helper validator:

python3 scripts/validate_mew_design_key.py --api-key "USER_PROVIDED_KEY"

Validation rule:

  • Exit code 0: key looks valid.
  • Exit code 2: auth failed or key is unusable.
  • Exit code 1: network or unexpected error.

This validator intentionally sends an invalid body to the Mew gateway so the response reveals whether authentication passed. Treat C40001 as a successful key validation signal. Treat C40100 to C40103 as failed validation.

Summarization Rules

  • Preserve meaning, not chronology.
  • Prefer outcomes over process.
  • Keep names, metrics, file paths, and decisions when they matter.
  • Omit apologies, greetings, retries, and duplicated wording.
  • If the thread is mostly open questions, frame the image around Current Status, Open Questions, and Suggested Next Steps.
  • If the thread ends with delivered work, frame the image around What Was Done, Files/Areas Touched, and What To Check Next.

Visual Rules

  • Favor a vertical summary card unless the user requests another format.
  • Ask for strong hierarchy, readable typography, and generous spacing.
  • Make the image feel like an intentional recap board, not a generic AI poster.
  • Keep copy density moderate so the result stays legible on chat screens.
  • If the user gives a brand color or product context, include it in the style direction.
  • Always add a subtle Mew.Design watermark at the bottom of the image. Keep it small, clean, and low-interference, but clearly readable.

Style Guide

  • minimalism

Use a structured card layout with large white space, sans-serif typography such as PingFang or Source Han Sans, and low-saturation Morandi accents like gray-blue or light beige. Good for workplace updates, project syncs, and formal consulting summaries. Must-see traits: large blank margins, restrained palette, clean card separation, minimal decoration, calm editorial rhythm. Avoid: shiny highlights, cartoon illustrations, fake 3D icons, dense bubble layouts.

  • healing-illustration

Use warm paper-like texture, gentle healing illustration cues such as plants, coffee cups, or simple characters, and a softer note-board composition. Good for community, customer care, and reflective personal summaries. Must-see traits: warm textured background, visible healing-style illustration elements, note or sticky-paper feeling, softer emotional atmosphere. Avoid: hard-edged enterprise UI, excessive glass effects, cold gradients, overly technical layout.

  • glassmorphism

Use blurred translucent panels, rounded floating containers, refined gradients, and bright 3D-like icon accents. Good for tech brands, premium communities, and polished knowledge-sharing content. Must-see traits: frosted glass panels, background blur through translucent cards, floating layered UI depth, visible edge highlights, soft shadows. Avoid: flat white cards, plain dashboard look, paper texture, chat screenshot aesthetics.

  • chat-bubbles

Use beautified IM-inspired dialogue bubbles, pastel or macaron accents, extracted quotes, and a final consensus card at the bottom. Good for interviews, testimonial-style content, and collaboration recaps. Must-see traits: obvious left-right conversation bubbles, quotable lines, chat rhythm, a final consensus block at the bottom. Avoid: generic report cards, glass dashboard cards, sparse poster layouts with no dialogue feel.

Style Locking

  • Treat the chosen style as a hard visual constraint, not a soft mood hint.
  • Make the chosen style's signature features explicit in the prompt.
  • Also mention what the image should not look like when that helps prevent drift.
  • If the first result only weakly resembles the selected style, retry once with stronger style language before returning it.
  • When retrying, say directly that the previous result drifted and specify the missing visual evidence.

Build The Payload

Use the helper script:

python3 scripts/build_summary_card_request.py \
  --title "本周 Agent 协作成果" \
  --summary "完成技能搭建,并把接口调用链路整理成可复用工作流。" \
  --point "新增两个可直接复用的 Codex skill" \
  --point "补齐了 API 调用脚本与参考文档" \
  --point "验证了脚本帮助信息与基本语法" \
  --next-step "真实联调一次图片生成接口" \
  --style minimalism \
  --output /tmp/chat-summary-body.json

The helper script already injects the Mew.Design bottom watermark requirement. Do not remove it unless the user explicitly asks to omit the watermark. The helper script should also inject stronger style-lock instructions so the output more faithfully matches the selected visual category, including must-see traits and avoid-list traits for every style.

Then generate the image through the official Mew design API:

curl -sS -X POST "https://api.mew.design/open/api/design/generate" \
  -H "Content-Type: application/json" \
  -H "x-api-key: USER_PROVIDED_KEY" \
  --data @/tmp/chat-summary-body.json

Do not run this generation step until the API key has already passed validation.

Style Selection

  • Use minimalism when the user wants clarity, professionalism, and structured information.
  • Use healing-illustration when the user wants warmth, softness, or emotional resonance.
  • Use glassmorphism when the user wants a modern, premium, technology-forward look.
  • Use chat-bubbles when the user wants the summary to retain conversation feel and quotable moments.
  • If the user does not choose after being prompted, default to minimalism.

Output Standard

When the image is generated successfully, respond with:

![Conversation summary](https://...)

[Open original image](https://...)

Add one short line that says what the image covers.

If the generated image visibly misses the watermark, retry once with a stronger prompt that explicitly says Place the text watermark "Mew.Design" at the bottom center of the image. If the generated image visibly misses the requested style, retry once with a stronger style-specific prompt that explicitly names the missing traits, such as strong frosted glass panels, warm paper texture with healing illustration, large white space and Morandi palette, or clear left-right chat bubbles.

Resources

平台分布

OpenClaw

73.19%
按下载量换算484

安全审计

VirusTotal

通过

ClawScan

通过

Static analysis

可疑

权限和风险

接触敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

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。来源安全扫描存在提醒或未通过项,安装前请自行复核。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills