# nvda-mcp

Source: https://glimind.com/server/pypi%2Fnvda-mcp
Registry id: `pypi/nvda-mcp`
Status: unknown

## Install

```json
{
  "mcpServers": {
    "nvda-mcp": {
      "command": "uvx",
      "args": [
        "nvda-mcp"
      ]
    }
  }
}
```

Claude Code:

```sh
claude mcp add nvda-mcp -- uvx nvda-mcp
```

## Tools (48)

- `nvda_access_connect`: Help someone via MCP Access: connect to their NVDA with code.
- `nvda_access_disconnect`: End MCP Access session, tools work on this computer again.
- `nvda_access_messages`: Read messages from the person you help, waiting for answer if asked.
- `nvda_access_send_message`: Say a message on the person's computer; NVDA keeps it for replay.
- `nvda_access_set_mode`: Let assistant act on their computer or only read and talk.
- `nvda_activate_object`: Press, follow or toggle a control found with nvda_find_objects.
- `nvda_activate_window`: Bring a window to front by title or handle.
- `nvda_audit_focus_order`: Tab through window, report unlabeled controls, silent controls, etc.
- `nvda_capture_control`: Picture of control on screen next to NVDA's description for it.
- `nvda_find_html`: Find web page elements by CSS selector via remote debugging.
- `nvda_find_in_page`: Move browse mode cursor to a match on the page.
- `nvda_find_objects`: Find controls by name and/or role in foreground window or background.
- `nvda_focus_object`: Move focus to a control found with nvda_find_objects.
- `nvda_get_action_log`: Get newest entries of the log of actions clients took.
- `nvda_get_addon_paths`: Get add-on code paths, scratchpad status, and running add-ons.
- `nvda_get_dev_info`: Get developer info for a control like NVDA+F1.
- `nvda_get_focus`: Returns the focused control: name, role, states, value, position, containers.
- `nvda_get_log`: Get NVDA's log filtered by level and text; IO entries optional.
- `nvda_get_navigator_object`: Returns NVDA's navigator object details.
- `nvda_get_object_tree`: Returns accessibility tree of foreground window, focus, navigator, or found control.
- `nvda_get_output_history`: Get beeps, sounds, braille output, speech cut off, merged optionally.
- `nvda_get_settings`: Read NVDA's current settings; secret-like values hidden.
- `nvda_get_speech_history`: Get what NVDA said with ids for polling and text filter.
- `nvda_get_status`: Returns NVDA version, synthesizer, speech mode, foreground app, permissions.
- `nvda_get_text`: Get full text of document, web page, edit field, or caret line.
- `nvda_inspect_html`: Get HTML element behind control in Chrome/Edge via remote debugging.
- `nvda_inspect_object`: Get IAccessible2/UIA attributes, containers, children, virtual list rows.
- `nvda_list_elements`: List headings, links, landmarks, form fields like Elements List.
- `nvda_list_scripts`: List NVDA commands available where focus is, with keys.
- `nvda_list_windows`: List open application windows, including on other virtual desktops.
- `nvda_lookup_api`: Check name in NVDA's Python code: existence, signature, docstring, members.
- `nvda_mouse_click`: Click, right-click or double-click a control, point, window or screen.
- `nvda_mouse_drag`: Drag from one point to another, e.g., a fader on canvas.
- `nvda_mouse_hover`: Move mouse over something to show tooltip and say what's under it.
- `nvda_mouse_scroll`: Turn mouse wheel over a point to scroll or turn knob.
- `nvda_move_to_element`: Move to or activate an element from nvda_list_elements.
- `nvda_press_keys`: Press keys with optional modifiers, report speech and focus.
- `nvda_preview_speech`: Preview speech without moving focus; scan window for unlabeled controls.
- `nvda_recognize_text`: Read screen text via OCR in a window or control.
- `nvda_reload_plugins`: Reload add-ons and app modules from disk like NVDA+Ctrl+F3.
- `nvda_run_scenario`: Run keys/typing, record speech after each step, compare with baseline.
- `nvda_run_script`: Run an NVDA command by name, whatever key it's assigned to.
- `nvda_set_document_mode`: Switch between browse mode and focus mode in web page.
- `nvda_set_value`: Experimental: set edit field text via accessibility API in background window.
- `nvda_speak`: Say something through NVDA's speech and braille.
- `nvda_trace_events`: Record NVDA events (focus, name/value changes, live regions) and speech.
- `nvda_type_text`: Type a line of text in any language without clipboard.
- `nvda_wait_for_speech`: Wait until NVDA says something, such as a chat reply.

Measured by Glimind (https://glimind.com): MCP handshake + tool list only, no tool is ever called. CC BY 4.0.
