Skip to main content
Glama
README.md
<div align="center">

<img src="docs/images/cover-en.png" alt="WorkBuddy-Link — one chat, an extra AI teammate" width="100%">

# WorkBuddy-Link

**Let WorkBuddy do the work. Stay in your AI assistant.**

[![Tests](https://github.com/P-PPPP/WorkBuddy-Link/actions/workflows/test.yml/badge.svg)](https://github.com/P-PPPP/WorkBuddy-Link/actions/workflows/test.yml)
[![Windows](https://img.shields.io/badge/Windows-0078D4?style=flat)](#-before-you-start)
[![Node.js](https://img.shields.io/badge/Node.js-24%2B-43853D?style=flat)](https://nodejs.org/)
[![License: MIT](https://img.shields.io/badge/License-MIT-8B5CF6?style=flat)](LICENSE)

[English](README.md) · [简体中文](README.zh-CN.md)

</div>

Ask your AI assistant to delegate a task to WorkBuddy, pass along the relevant conversation context, and work in the same project folder. Get charts, models, and other results back in your current chat.

**The connector is free and needs no separate API key.** Tasks use your WorkBuddy account; its limits and any model charges still apply.

## ✨ What you get

| | What it means for you |
|:---:|---|
| 💬 **One conversation** | Delegate work and receive results without switching to another chat. |
| 📂 **Shared project files** | Pass the task brief, agreed requirements, and project folder path together. |
| 🛠️ **Your existing tools** | Use the MCP tools and skills you have configured in WorkBuddy. |
| 🎛️ **Your choice of model** | Choose the execution model, check progress, or stop a task. |

## 🚀 Get started

Copy this and send it to your AI assistant:

```text
Install WorkBuddy-Link for me: https://github.com/P-PPPP/WorkBuddy-Link
```

### 📋 Before you start

- A Windows computer with WorkBuddy installed and signed in.
- [Node.js 24+](https://nodejs.org/) and Git.
- An AI assistant that supports local MCP servers and can run the installation steps on your computer.
- For drawing or modeling: the corresponding tool must already work in WorkBuddy.

<details>
<summary>Prefer to install it manually?</summary>

Open PowerShell in the folder where you want to keep the project:

```powershell
git clone https://github.com/P-PPPP/WorkBuddy-Link.git
Set-Location WorkBuddy-Link
npm ci --ignore-scripts --no-audit --no-fund
npm run smoke
npm run configure
```

Copy the generated `workbuddy-link` entry into your AI assistant's MCP settings, keeping any existing servers. Reconnect the assistant, then ask it to check WorkBuddy's status. The smoke check does not call a model.

[Detailed setup and settings](docs/REFERENCE.md#installation)

</details>

## 💬 Just ask your assistant

Start with your current project:

> Give this task to WorkBuddy. Share the relevant context and our project folder path. Ask me which model to use first, then save the results in that folder and report back here.

Your assistant passes the context it selects; the full chat history is not synced automatically.

### 📈 Draw a chart in Origin

> Have WorkBuddy use Origin MCP to plot this data. Include the file paths, labels, and style we agreed on. Ask me which model to use, then save PNG and PDF files in our project folder.

The chart below was created through WorkBuddy using OriginPro 2024 and synthetic sine/cosine data. [Test notes](docs/VALIDATION.md#origin-mcp-demonstration)

![Chart created by WorkBuddy through Origin MCP](docs/images/origin-demo.png)

### 🧊 Make a model in Blender

> Have WorkBuddy use Blender MCP to make a desk organizer using the dimensions and design we discussed. Ask me which model to use, then return the .blend file and a preview image in our project folder.

This is a suggested workflow; Blender has not been tested in this project.

### ⏯️ Check in or change direction

- “Which tools and skills can WorkBuddy use?”
- “Check the progress of my WorkBuddy task.”
- “Stop the current WorkBuddy task.”

## ❓ Quick answers

| Question | Answer |
|---|---|
| **Does it control my open WorkBuddy chat?** | It starts a separate task session. Do not expect it to appear in the desktop chat history. |
| **Which model does it use?** | Ask your assistant to let you choose first. Otherwise WorkBuddy uses its default, which may be Auto. |
| **My assistant cannot connect.** | Check Node.js 24+, reconnect the MCP server, and run `npm run smoke` in this project. |
| **WorkBuddy asks me to sign in.** | Sign in through the desktop app, then start a new task session. |

[More settings and troubleshooting](docs/REFERENCE.md) · [Report an issue](https://github.com/P-PPPP/WorkBuddy-Link/issues)

## 🔐 Privacy & compatibility

WorkBuddy-Link uses your local WorkBuddy sign-in. Task records stay on your computer; WorkBuddy sends the task and any tool-provided content to the services it uses. Only delegate files and tasks you intend it to access. Do not upload local account files or the `data` folder.

This is an unofficial community project, tested with **WorkBuddy 5.9.3 on Windows**. WorkBuddy updates may require compatibility fixes.

## 📄 License

[MIT](LICENSE). Personal learning and non-commercial use are encouraged as a voluntary preference; commercial use is permitted by the license.

TDQS

A3.5/5.0

Scored across 7 tools

Disambiguation4/5

Most tools have clearly distinct roles: send_message starts or continues work, stop_task cancels, and the select_* tools configure model or MCP/skills. However, get_message, wait_for_message, and get_status overlap somewhat in observing task events and pending input, which could require careful reading of descriptions to choose correctly.

Naming Consistency5/5

All seven tools follow a consistent snake_case verb_noun pattern such as send_message, get_status, stop_task, and select_model. The naming is predictable and easy to map to actions.

Tool Count5/5

Seven tools is well-scoped for an asynchronous worker interface covering task messaging, observation, cancellation, and configuration. Each tool appears to earn its place without obvious redundancy.

Completeness5/5

The surface covers the full interaction lifecycle: starting/continuing tasks, reading incremental events, waiting for events or input, checking health/status, cancelling tasks, and selecting runtime model and MCP/skill configuration. No obvious operational gaps are apparent for a local async worker.

Maintenance

ActivityMaintained
ResponsivenessSyncing