yyq2024

Draw.io Skill

设计创意v1.0.1
作者:yyq2024更新于 2026-05-11共 2 个版本 · 最新版本发布于 2026-03-20drawio-skill@clawhub_yyq2024/drawio-skill
视觉素材

据 AlphaMoat 统计,截至 2026-10,Draw.io Skill 累计下载 9,370 次、获 0 个星标,按累计下载量在本次参与排名的 10,914 个 AI 技能中排名第 2,944 位,在「设计创意」分类(1,116 个)中排名第 382 位。

列表默认按累计下载量排序,也可按星标排序;详情页排名按累计下载量计算。

累计下载
9,370
星标
-
全站下载量排名
#2,944
全站前 27% · 「设计创意」前 35%
平台累计下载星标评论
ClawHub ↗4,311-1
SkillHub ↗9,370-—

各平台的计数口径不同,统计时间不一致。

这个技能是做什么的

该技能要求先读取当前图表再编辑,并可导出可编辑源文件或用于分享的 SVG。技能卡提示推荐的第三方 MCP 服务包未固定版本,操作可能修改实时图表或在工作区写入导出文件,且依赖外部 MCP 连接及当前会话状态。安装前审查固定版本;编辑前读取现有内容并保留用户变更,导出后检查文件。 新建工作或在新连接中检查前先启动 session,编辑前读取真实 cell ID。技能本身不打包或安装 MCP 服务,工具缺失时应重新连接并确认可用性。

资料发布于 2026-10-29

可以用来做什么

  • 创建流程图、架构图或机器学习图表
  • 读取并编辑当前 draw.io 图表
  • 导出可编辑文件或 SVG

什么时候用

  • drawio MCP 服务已配置且当前会话状态可用时
  • 确认将修改的图表内容,并检查依赖包版本和导出目标后

示例提示词

以下是根据技能文档生成的示例,不代表实际运行结果。

先读取当前图表并说明将修改的内容;在我确认前不要编辑或导出文件。

月度数据趋势

累计下载9,370
05,0521.0万2026-03 · 累计下载 1912026-05 · 累计下载 1,2582026-06 · 累计下载 1,7842026-07 · 累计下载 2,9922026-08 · 累计下载 3,8182026-09 · 累计下载 5,0602026-10 · 累计下载 9,37026-0326-0926-10
星标-
0112026-03 · 星标 -2026-05 · 星标 -2026-06 · 星标 -2026-07 · 星标 -2026-08 · 星标 -2026-09 · 星标 -2026-10 · 星标 -26-0326-0926-10

2026-09 起统计方法调整:同名技能按作者分别统计,并覆盖全部技能。图中虚线为调整前的月份,与之后的数据不直接比较,也不计算跨越调整的增长。

月份累计下载环比星标
2026-109,370—-
2026-095,060—-
2026-083,818+27.6%-
2026-072,992+67.7%-
2026-061,784+41.8%-
2026-051,258—-
2026-03191—-

SKILL.md 原文

以下是作者发布的原始文件,按纯文本显示,本站未做修改或审核;其中的链接不可点击,命令请核实后再运行。抓取于 2026-10-23。

5.3 KB · ClawHub v1.0.1
---
name: drawio-skill
description: Create and edit draw.io diagrams through the configured drawio MCP server, including flowcharts, architecture diagrams, ML model diagrams, Chinese labels, animated connectors, and svg export.
---

# Draw.io Skill

Use this skill for all draw.io and diagrams.net work. This skill wraps the configured `drawio` MCP server; it does not replace the server.

## Required workflow
展开全部(共 112 行)

1. Start the draw.io session first.
- Call `start_session` before creating a new diagram or before inspecting or editing a diagram in a fresh session.

2. Create new diagrams from complete XML.
- Use `create_new_diagram` with a full `mxGraphModel`.
- Keep the diagram within a single viewport when practical.

3. Read before editing.
- Always call `get_diagram` before `edit_diagram`.
- Do not guess cell IDs.
- Preserve user changes already present in the live session.

4. Export deliverables.
- Export `.drawio` for editable source.
- Export `.svg` when the user needs a shareable artifact or when the diagram uses animated connectors.

5. Report outputs clearly.
- Tell the user which files were exported and where they were written.

## Export naming convention

- Write exports to the current workspace unless the user specifies a path.
- Derive a short kebab-case basename from the main topic when possible, such as `user-login-flow`, `payment-architecture`, or `event-pipeline`.
- If the topic is unclear, use `diagram`.
- For Chinese-language outputs, prefer an ASCII basename with a `-zh` suffix, such as `order-flow-zh`.
- For architecture diagrams, prefer `-architecture`, `-topology`, or `-flow` suffixes when they improve clarity.
- For machine-learning diagrams, prefer model-aware basenames such as `lstm-architecture`, `transformer-blocks`, or `cnn-pipeline`.
- When exporting both editable and shareable artifacts, keep the same basename for both, such as `diagram.drawio` and `diagram.svg`.
- Avoid spaces and non-ASCII characters in filenames unless the user explicitly requests them.

## Language and labeling rules

- If the request is primarily in Simplified Chinese, reply in Simplified Chinese.
- If the user wants Chinese labels, legends, or summaries, keep the diagram wording in Simplified Chinese.
- Keep standard abbreviations and proper nouns in their common form, such as API, Kafka, Redis, Kubernetes, LSTM, CNN, and Transformer.
- Keep node text short enough to fit inside shapes.

## General diagram rules

- Prefer clear left-to-right or top-to-bottom flow.
- Use explicit edge routing and waypoints when edges would otherwise overlap.
- Use titles, legends, and note boxes sparingly.
- Use color intentionally to separate data flow, control flow, or state.

## Architecture-specific rules

- Identify the major layers first, such as clients, edge, services, data, and external systems.
- Draw boundaries for environments or trust zones when relevant, such as browser, VPC, cluster, region, or third-party systems.
- Keep arrows directional and semantically consistent.
- Label protocols, topics, or data contracts only when they matter to the reader.
- Use containers for service groups, domains, or infrastructure groups instead of loose boxes.

## ML-diagram rules

- Show the main flow clearly from input to output.
- Use repeated blocks, grouped containers, and stage labels for model structure.
- Label tensor shape, sequence length, channel count, or embedding size only when it helps understanding.
- Use consistent color semantics for inputs, hidden states, attention paths, residual paths, and outputs.
- For recurrent or temporal models, make state flow visually distinct from data flow.
- For LSTM and RNN diagrams, separate cell state and hidden state.
- For Transformer diagrams, distinguish embeddings, attention, feed-forward blocks, residual paths, and normalization.
- For CNN diagrams, emphasize feature-map progression, pooling, skip paths, and classifier head.

## Animated connectors

When the user asks for animated connectors, flowing lines, or moving dashed lines:

- Add edge styling such as `dashed=1;dashPattern=8 4;flowAnimation=1;`.
- Keep arrowheads explicit with `endArrow=classic` unless the requested style suggests otherwise.
- Export to `.svg` because the animation is preserved there.

## Setup and recovery

This skill expects the `drawio` MCP server to already be configured and available.

- If the server is unavailable, tell the user that the drawio MCP connection needs to be configured or reconnected before continuing.
- After the connection is restored, restart the Codex session or reload the IDE window if the tools still do not appear.

## Good triggers

- "Create a system diagram in draw.io"
- "Create a Chinese-labeled draw.io diagram"
- "Edit this flowchart"
- "Export the current diagram as draw.io and svg"
- "Create a diagram with animated connectors"
- "Turn this process into a draw.io diagram"
- "Create a cloud architecture diagram"
- "Map a deployment topology"
- "Draw a business process flowchart"
- "Create an animated LSTM diagram"
- "Draw a Transformer architecture"
- "Create a CNN block diagram"
- "Update the existing draw.io layout and keep the same structure"
- "Generate a box-and-arrow diagram for this workflow"

## Limitations

- This skill depends on a working `drawio` MCP server.
- The editing workflow operates on the live draw.io session state exposed by the MCP tools.
- This skill does not bundle or install the MCP server by itself.

「设计创意」分类下的热门技能

视频短视频图片音乐音频配音口播数字人全能创作设计创意

通过同一个AI创作工具完成AI图片、视频、音乐与语音创作,查找公开社交数据,编辑视觉成果,并集中查看和管理生成资产。Beatra 全能创作支持AI图片生成(文生图)、AI视频生成(图生视频)和AI视频编辑,通过AI音乐生成把想法做成歌曲或纯音乐,也可用AI语音生成(文字转语音)、多语言旁白和AI语音克隆完成跨媒介内容;还能查找抖音数据、小红书数据,以及 YouTube、TikTok、Instagram 与 X 上的公开帖子、笔记、评论、账号和趋势,把社交媒体数据用于竞品调研和创作。它是一款面向AI内容创作与多模态内容创作的AI创作套件,用起来就是一个把图片、视频、音乐与语音收在同一处的AI创作助手,适用于社交内容、产品展示、营销素材、课程、播客、短片和跨媒介项目;还能在不同媒介间复用原始素材、已有成果和查到的公开资料,查看制作进度、返回文件、实际用量和积分记录。

↓ 147.6万★ 113u_319b171b
AI数字人口播与虚拟主播视频设计创意

用一张人像和一段短脚本或语音,制作有明确讲述内容的数字人口播视频。可使用现成录音,也可以先选择音色,把脚本生成旁白,让画面中的人物随声音完成稳定、自然的口播。适合 AI 主播、虚拟主播视频、数字人讲解、让照片说话、照片说话视频、产品讲解视频、数字人培训视频、公告和社交口播,并在交付后检查人物一致性、声音清晰度、口型同步和动作稳定性。一节课的讲解也能做成课程口播。

↓ 50.8万★ 39u_319b171b
架构图一键生成设计创意

ContextWeave 架构图一键生成

↓ 47.0万★ 280user_bddf3fe6
海报设计工作室设计创意

将一段活动描述、一张产品照片或品牌参考图,转化为抓人眼球的活动海报、宣传 banner 或社交媒体配图。这款 AI海报设计与制作工具生成图片丰富的海报,在海报排版上搭配清晰视觉层级、按品类匹配的风格和预留文字区域,覆盖活动、音乐节、电影、产品发布和促销场景。你可以把它当作 AI宣传单制作工具来制作促销宣传单,当作活动海报设计工具,当作电影海报生成器,当作音乐节海报设计工具,当作产品发布会海报的 banner设计 AI,或当作平面设计海报工作室。从活动主题出发,用品牌风格参考图变换产品或场景照片,或在已认可草稿上做定向精修,完成可印刷或可分享的成品。

↓ 30.5万★ 28user_87b8e34f
自动化克隆数字人口播混剪视频设计创意

抖音短视频双路径生产体系(单 Skill 版)。Path A[付费·高质量]:策略大脑→(可选:素材混剪/AI高光提取)→脚本优化→AI生图/素材准备→配音渲染(百炼/GPT-SoVITS/CosyVoice/VoxCPM2+剪映/Remotion/ffmpeg/数字人)。Path B[免费·动画风]:采集热点→脚本→edge-tts免费配音+HyperFrames开源渲染→MP4,零云费开箱快。触发词分层:①"抖音短视频/选题规划/合规闸门/运营策略" ②"口播脚本/AI生图/AI配音/视频渲染/Remotion渲染" ③"蹭热点/快速出片/免费做视频" ④"混剪/高光提取/素材粗剪/autoclip

↓ 29.7万★ 79user_3b54b400
GPT Image 2.5生图设计创意

GPT Image 2.5生图是一款 AI 工具,用 GPT-Image-2.5 和 ChatGPT Images 2.5 出一张图。GPT Image 2.5 Studio 支持 gpt-image-2.5-flare 与 gpt-image-2.5-sunburst,可文生图或参考图生图,做中文海报、商品图、信息图和社媒画面。

↓ 27.1万★ 27user_87b8e34f
MiniMax H3 AI视频创作设计创意

用 MiniMax H3 这款 AI视频生成器创作高质感 2K AI 视频。从文字创意出发完成文生视频,让静态图片自然动起来,用首尾画面做首尾帧视频的完整转场,也可以结合图片、视频和音频参考,以多模态视频生成塑造统一场景。适合 AI 广告视频、电商产品视频、品牌宣传视频、动态海报、游戏 UI、电影片头和社交短视频创作,形成电影感画面与同步的原生立体声。使用 Beatra 账号即可开始,创作进度与成片集中管理。

↓ 26.3万★ 23user_87b8e34f
AI Logo设计·品牌标识创作设计创意

输入品牌名称、行业方向或参考图,生成专业 AI Logo、品牌标识和应用图标。这款 AI logo 设计工具从你的品牌简报出发——公司名、行业气质、配色偏好——先探索多个商标设计方向,再用精确品牌色、简洁几何和可缩放构图精修最强概念。制作企业 logo、创业公司标志、字母标、徽章、组合标识和 VI设计中的主标,从网站 favicon 到门头招牌都清晰可辨。从文字描述出发,将已有草图和品牌素材转化为精致标识,或在已认可草稿上做定向精修。

↓ 26.0万★ 19user_87b8e34f

数据月份:2026-10 · 下载/星标数聚合自公开技能注册库(ClawHub、SkillHub),按月更新。口径说明见数据方法。