openworkbuddy

Name: openworkbuddy
Author: CatCatUncle

Warning

Local-first AI office agent that hands you files, not chat logs: swap any LLM (or use local Claude Code/Codex), keep data on your machine, add a skill with one Markdown file. 本机 AI 办公 agent：吐文件不吐聊天记录，不绑模型，数据自己留着。

54stars

17forks

JavaScript

Installation

# Add to your Claude Code skills
git clone https://github.com/CatCatUncle/openworkbuddy

Getting Started

Guides for using ai agents skills like openworkbuddy.

Caveman: Cut Claude Token Use by 65%
How agent-side prompt compression works, when to use it, and when not to.
What is an AI Skills Marketplace?
Definitions, how marketplaces work, and how to choose between them in 2026.
Getting Started with AI Skills

Security ReportWarning

Last scanned: 9/9/2026

{
  "issues": [
    {
      "type": "npm-audit",
      "message": "body-parser: Vulnerability found",
      "severity": "medium"
    },
    {
      "type": "npm-audit",
      "message": "exceljs: Vulnerability found",
      "severity": "medium"
    },
    {
      "type": "npm-audit",
      "message": "express: Vulnerability found",
      "severity": "medium"
    },
    {
      "type": "npm-audit",
      "message": "image-size: image-size: ICNS parser allows denial of service through an infinite loop",
      "severity": "high"
    },
    {
      "type": "npm-audit",
      "message": "pptxgenjs: Vulnerability found",
      "severity": "high"
    },
    {
      "type": "npm-audit",
      "message": "qs: qs array-limit bypass via bracket-key comma parsing",
      "severity": "medium"
    },
    {
      "type": "npm-audit",
      "message": "uuid: uuid: Missing buffer bounds check in v3/v5/v6 when buf is provided",
      "severity": "medium"
    }
  ],
  "status": "WARNING",
  "scannedAt": "2026-09-09T08:41:35.317Z",
  "npmAuditRan": true,
  "pipAuditRan": true,
  "promptInjectionRan": true
}

README.md

Frequently Asked Questions

What is openworkbuddy?

openworkbuddy is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by CatCatUncle. Local-first AI office agent that hands you files, not chat logs: swap any LLM (or use local Claude Code/Codex), keep data on your machine, add a skill with one Markdown file. 本机 AI 办公 agent：吐文件不吐聊天记录，不绑模型，数据自己留着。. It has 54 GitHub stars.

Is openworkbuddy safe to use?

openworkbuddy returned warnings in SkillsLLM's automated security scan. It has no critical vulnerabilities, but review the flagged issues in the Security Report section before adding it to your workflow.

How do I install openworkbuddy?

Clone the repository with "git clone https://github.com/CatCatUncle/openworkbuddy" and add it to your Claude Code skills directory (see the Installation section above).

What programming language is openworkbuddy written in?

openworkbuddy is primarily written in JavaScript. It is open-source under CatCatUncle on GitHub, so you can review or fork the full source.

Are there alternatives to openworkbuddy?

Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh openworkbuddy against similar tools.

Agentic AI for Beginners

Build your first AI agent from scratch - tool use, ReAct pattern, memory, deployment

41 minBeginner

Comments (0)

to leave a comment.

No comments yet. Be the first to share your thoughts!

Related Skills

ECC

by affaan-m

The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.

254,598

mediagen phoenix

⚠️ Third-Party Software Notice

This skill is third-party open-source software developed and hosted independently on GitHub. SkillsLLM is an informational directory and does not control or maintain the underlying repository.

Any security checks, ratings, or warnings displayed by SkillsLLM are automated and limited in scope. They do not constitute a security certification or guarantee that the software is safe, error-free, or free from malicious code, vulnerabilities, compromised dependencies, or prompt-injection risks.

Review the source code, permissions, dependencies, and configuration before installing or running any third-party skill. Use is at your own risk. To the maximum extent permitted by applicable law, SkillsLLM is not liable for losses arising from third-party software.

Read the Terms of Service

为什么是它

交付的是文件，不是聊天记录。 PPT / Word / Excel / 网页都是真生成的，成果面板里点开就能验收。声称写了文件却不在磁盘上，会被当场拦下重做。

不绑任何一家模型，东西都在你手里。 DeepSeek / 通义 / 智谱 / Kimi / OpenRouter / Ollama 本地模型界面点一下就切；本机装了 Claude Code / Codex 的，一键拿它当发动机，不再另买 token。自托管，会话、文件、Key 全在本机，默认只监听 127.0.0.1。

加一个能力 = 丢一个 Markdown 文件。 放进 skills/，存盘后下一条任务就生效——不改代码、不重启、不打包。往外接 MCP 连接器和 Agent Plugins 开放标准，别人的插件粘个 GitHub 地址就装。

它替你做完的事

你说一句	它交给你
帮我出一份 Q3 复盘 PPT，数据用这个 Excel	读表 → 算 → 一个能直接放的 `.pptx`
调研国内 AI 陪伴产品，出一份报告	联网搜 → 逐个打开读 → Markdown / Word
把这份材料做成手机上能看的网页	写 HTML → 起本机服务 → 扫码就能看
每天 9 点抓行业新闻，做成晨报发我飞书	定时任务 + IM 推送，错过了会补跑

还有多任务并行、Goal 目标验收、👍👎 反馈进自进化、双层记忆、权限档位、IM 远程指挥、桌面宠物……全部能力见 功能清单。

跑起来

装包：去 Releases 下对应的包，打开后五步向导带你注册账号、粘 Key（当场验活）、选引擎——三分钟内说出第一句话。

系统	下哪个
macOS · Apple 芯片	`OpenWorkBuddy-*-mac-arm64.dmg`
macOS · Intel	`OpenWorkBuddy-*-mac-x64.dmg`
Windows 10/11（x64 与 ARM 同一个包，装时自动选）	`OpenWorkBuddy-*-win-setup.exe`
Windows 免安装版（U 盘/公司电脑不让装软件）	`OpenWorkBuddy-*-win-x64-portable.exe` / `-win-arm64-portable.exe`（不知道自己是哪种 CPU 就拿 `-win-portable.exe`，两种都装在里面，体积翻倍）

macOS 第一次打开提示「无法验证开发者」——包没买苹果证书签名，不是有毒。右键 → 打开，或 xattr -cr /Applications/OpenWorkBuddy.app。你的数据在 ~/OpenWorkBuddy，卸载不会删。

从源码（Node.js 18+，零构建零框架，改完刷新即生效）：

git clone https://github.com/CatCatUncle/openworkbuddy.git
cd openworkbuddy && npm install
npm run app     # 桌面版；或 npm start 走浏览器 http://localhost:3800

然后在输入框里说人话，比如「帮我做一份介绍 OpenWorkBuddy 的 PPT」。看着不顺眼就改：头像菜单 → 外观，主题 / 六套皮肤 / 字号四档 / 字体 / 紧凑密度，点一下立刻生效。

镜像、一键脚本、端口占用、启动卡住 → 安装与启动

配模型

设置 → 模型，选渠道预设（OpenAI / Anthropic / OpenRouter / 火山方舟 / 百炼 / DeepSeek / 智谱 / Kimi / Ollama），地址和协议自动填好，只差粘 Key，保存即热生效。对照表见配置模型。带 reasoning 的模型能在界面里关掉思考或调档。

config.json 是唯一存着 API Key 的文件，已在 .gitignore 里，别手滑提交。

命令行也能用

wb 跟桌面版共用同一份配置、技能、记忆和连接器，答案落到 stdout，能塞进任何管道、脚本和 cron：

npm link                                   # 一次性：把 wb 装成全局命令
wb "帮我写一份本周周报"                     # 单发：跑完就退，退出码 0/1 说实话
cat error.log | wb "这是什么问题"           # 管道：内容当附加材料一起送进去
wb --json "整理会议纪要" | jq -j 'select(.type=="text") | .delta'   # NDJSON 事件流给脚本用
wb engines && wb engines use claude-code   # 本机装了 Claude Code / Codex？一键拿它当底层

-q 只要答案、-c 续上一轮、-C <目录> 指定工作目录、wb sessions 列会话。全部参数 → 命令行用法。

⚠️ 这个 agent 手里有 shell

它能执行命令、读写文件、访问网络——所以闸门是真拦的：命令审批、文件黑名单、URL 白名单、审计日志、四档权限。放到公网前务必先读安全，默认配置只为本机使用而调。

交流群

用崩了、有想法、想一起改，进飞书群直接说。飞书扫码：

一起把它做下去

用崩了、卡住了，开个 issue —— 哪怕只贴一句报错。 你以为「只有我遇到」的坑，多半所有人都在踩。贴之前扫一眼，别把 API Key 带上。

想动手，按投入从小到大三条路：

10 分钟 —— 写个技能。一个 Markdown 文件丢进 skills/，存盘即生效，三分钟模板在这
1 小时 —— 补一个模型渠道预设、修一处文档、给某个文件格式加上预览
一晚上 —— 挑个 issue 改。npm install && npm start 就跑起来，npm test 不需要 API Key 就能全绿

项目结构、测试、PR 规范都在参与贡献。不用先开 issue 问，直接发 PR。

文档


功能清单	全部能力、内置技能与工具
安装与启动	安装包、源码、Windows、常见卡壳
配置模型	各服务商 base_url / 模型名对照
命令行用法	CLI 参数、管道、`--json`、脚本和 cron
扩展	写技能、接 MCP、装插件、建专家
IM 与定时任务	飞书 / QQ / 企微 / 微信 / 钉钉，cron
多人协作	多租户、账号、权限、积分额度
安全	审批闸门、黑白名单、审计
部署	服务器 / Docker / 反代
实现细节	agent 主循环怎么转的
参与贡献	项目结构、测试、提 PR

协议

一句话：自己用、学习用、非营利机构用——免费；拿去赚钱（公司内部提效也算）——找作者买商业授权。 协议是 PolyForm Noncommercial 1.0.0，哪些算商用、怎么谈见 COMMERCIAL-LICENSE.md。

免责

本项目是对腾讯 WorkBuddy 产品形态的独立开源实现，与腾讯没有任何关系，不含其任何代码或资源。「WorkBuddy」是其权利人的商标。