# oswright MCP server

> Enables AI agents to automate desktop applications using incremental perception to reduce token usage by up to 7x.

- Repository: https://github.com/Ask-812/oswright
- Page: https://hostedmcp.org/servers/ask-812/oswright
- Transport: stdio
- Stars: 4
- License: MIT
- Latest version: 0.8.2

## Install

### Claude Code

```sh
claude mcp add oswright -- uvx oswright
```

### Cursor (`~/.cursor/mcp.json`)

```json
{
  "mcpServers": {
    "oswright": {
      "command": "uvx",
      "args": [
        "oswright"
      ]
    }
  }
}
```

### VS Code (`.vscode/mcp.json`)

```json
{
  "servers": {
    "oswright": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "oswright"
      ]
    }
  }
}
```

### VS Code CLI

```sh
code --add-mcp '{"name":"oswright","type":"stdio","command":"uvx","args":["oswright"]}'
```

### Claude Desktop (`claude_desktop_config.json`)

```json
{
  "mcpServers": {
    "oswright": {
      "command": "uvx",
      "args": [
        "oswright"
      ]
    }
  }
}
```

### Codex

```sh
codex mcp add oswright -- uvx oswright
```

### Codex config.toml (`~/.codex/config.toml`)

```toml
[mcp_servers.oswright]
command = "uvx"
args = ["oswright"]
```

### Gemini CLI

```sh
gemini mcp add -s user oswright uvx -- oswright
```

### Gemini settings.json (`~/.gemini/settings.json`)

```json
{
  "mcpServers": {
    "oswright": {
      "command": "uvx",
      "args": [
        "oswright"
      ]
    }
  }
}
```

## Tools (43)

- `screenshot`
- `get_screen_info`
- `find_text_on_screen`
- `read_screen_text`
- `find_image_on_screen`
- `mouse_click`
- `mouse_double_click`
- `mouse_move`
- `mouse_scroll`
- `mouse_drag`
- `get_mouse_position`
- `type_text`
- `press_key`
- `click_text`
- `double_click_text`
- `right_click_text`
- `hover_text`
- `fill_field`
- `fill_form`
- `wait_for_text`
- `wait_for_text_gone`
- `wait_for_time`
- `list_windows`
- `focus_window`
- `close_window`
- `minimize_window`
- `screenshot_window`
- `get_clipboard`
- `set_clipboard`
- `launch_app`
- `get_ocr_info`
- `observe`
- `find_element`
- `click_element`
- `read_model_text`
- `perception_stats`
- `remember_screen`
- `atlas_stats`
- `get_ui_tree`
- `click_ui_element`
- `fill_ui_element`
- `get_active_window`
- `wait_for_change`

## FAQ

### How does OSWright reduce token costs?

It uses incremental perception to rescan only the parts of the screen that changed and maintains a screen model to reuse previous observations.

### Which operating systems are supported?

OSWright is cross-platform, supporting Windows (via Win32 API), Linux (via pynput/X11), and macOS (via pynput/Quartz).

### Does it require a large installation on Windows?

No, on Windows it uses the built-in OCR engine, making the installation only a few MB instead of several GB.
