A desktop AI assistant for programming, file management, and any task you can describe.
Inspired by Anthropic's Cowork — run multiple AI tasks in parallel with a friendly GUI.
Fully compatible with Claude Code configuration, meaning you can use any Anthropic-compatible LLM.
Not just a GUI. A real AI collaboration partner. No need to learn the Claude Agent SDK — just describe tasks in natural language.
agent-cowork-demo.mp4
Claude Code is powerful — but it only runs in the terminal.
That means:
- No visual feedback for complex tasks
- Hard to track multiple sessions
- Tool outputs are inconvenient to inspect
Claude Cowork solves these:
- Runs as a native desktop application
- Acts as your AI collaboration partner
- Run up to 3 tasks in parallel
- Reuses your existing
~/.claude/settings.json - 100% compatible with Claude Code
If Claude Code works on your machine — Claude Cowork works too.
Queue multiple tasks and run them simultaneously:
- Up to 3 concurrent tasks running in parallel
- Quick task input with floating action button
- Real-time task status tracking
- Toast notifications when tasks complete
- Cancel tasks at any time
Manage MCP servers and skills through Settings:
- Built-in MCP Servers: Filesystem, Web Fetch, Memory
- Auto-discover skills from
~/.claude/skills/ - Toggle connectors and skills on/off
- Configure preferences (max tasks, auto-start, notifications)
See exactly what Claude is changing in your files with inline diffs:
- Edit operations show before/after comparisons
- Write operations show full file diffs
- Syntax highlighting and collapsible large diffs
Built-in dev-browser skill for web tasks:
- Navigate websites and take screenshots
- Test web applications
- Fill forms and interact with pages
- Welcome Screen for new users with quick task suggestions
- Progress Panel showing real-time activity and stats
- Simplified tool display with friendly descriptions
- Clean, modern interface
- Create sessions with custom working directories
- Resume any previous conversation
- Complete local session history (SQLite)
- Safe deletion and automatic persistence
- Token-by-token streaming output
- View Claude's reasoning process
- Markdown rendering with syntax-highlighted code
- Visualized tool calls with status indicators
- Explicit approval for sensitive actions
- Allow or deny per tool
- Interactive decision panels
- Full control over what Claude can do
- Claude Code installed and authenticated
- Node.js 18+ (or Bun)
# Clone the repository
git clone https://github.com/nicekid1/Claude-Cowork.git
cd claude-cowork
# Install dependencies
npm install
# Run in development mode
npm run dev
# Or build production binaries
npm run dist:mac # macOS
npm run dist:win # Windows
npm run dist:linux # LinuxClaude Cowork shares configuration with Claude Code.
It directly reuses:
~/.claude/settings.json
This means:
- Same API keys
- Same base URL
- Same models
- Same behavior
Configure Claude Code once — use it everywhere.
Access settings via the sidebar to configure:
| Setting | Description |
|---|---|
| Connectors | Enable/disable MCP servers |
| Skills | Toggle discovered skills |
| Max Concurrent Tasks | 1-5 parallel tasks |
| Auto-start Tasks | Automatically start queued tasks |
| Show Notifications | Toast alerts for task completion |
Claude Cowork auto-discovers skills from ~/.claude/skills/:
| Skill | Description |
|---|---|
| Dev Browser | Browser automation with Playwright |
| Office Documents | Create Word, Excel, PowerPoint files |
| Backend Guidelines | Electron/Node.js development patterns |
| Frontend Guidelines | React 19/TypeScript best practices |
| Error Tracking | Error handling patterns |
| Skill Developer | Create custom skills |
| Layer | Technology |
|---|---|
| Framework | Electron 39 |
| Frontend | React 19, Tailwind CSS 4 |
| State Management | Zustand |
| Database | better-sqlite3 (WAL mode) |
| AI | @anthropic-ai/claude-agent-sdk |
| Build | Vite, electron-builder |
# Start development server (hot reload)
npm run dev
# Type checking
npm run build
# Lint
npm run lintsrc/
├── ui/ # React 19 frontend
│ ├── components/ # UI components
│ ├── store/ # Zustand state
│ └── hooks/ # Custom hooks
├── electron/ # Electron main process
│ ├── libs/ # Core libraries
│ └── types.ts # Shared types
└── skills/ # Bundled skills
.claude/
├── skills/ # Auto-discovered skills
├── hooks/ # Automation hooks
└── settings.json # Claude Code config
- Parallel task queue (up to 3 concurrent)
- Settings panel for connectors & skills
- Welcome screen & progress panel
- Office document creation skill
- GUI-based configuration for models and API keys
- Multi-agent orchestration
- Autonomous checkpoints with git integration
- Project memory and context persistence
Pull requests are welcome.
- Fork this repository
- Create your feature branch
- Commit your changes
- Open a Pull Request
MIT
