Oswright

Get the current mouse cursor position.

Get information about the active OCR backend and available backends. Useful for debugging OCR issues.

Get screen dimensions and monitor count. Args: monitor: Monitor index (0 = all monitors combined).

Get the accessibility tree of the focused window (or a specific window). Returns all interactive UI elements with their names, types, and positions. This is deterministic and instant — much more reliable than OCR for apps with proper accessibility support (most modern Windows apps). Windows only. Falls back gracefully on Linux/macOS. Args: window_title: Optional window title to inspect (default: focused window). max_depth: How deep to walk the UI tree (default: 8).

Find text on screen using OCR and move the mouse over it (hover). Useful for triggering tooltips, hover menus, or inspecting elements. Returns action result and a screenshot of current screen state. Args: text: The visible text to find and hover over. exact: Require exact text match. timeout: Maximum seconds to wait for the text to appear. poll_interval: Seconds between each retry. monitor: Monitor index.

Launch an application. Optionally wait for specific text to appear on screen (indicating the app has loaded). Returns a screenshot after launch. The command is executed directly, never through a shell, so shell syntax (pipes, redirection, chaining) is not interpreted and not available. Args: command: Application name or full path (e.g. 'notepad', 'calc', r'C:\Windows\System32\notepad.exe'). args: Optional argument list. Prefer this over embedding arguments in `command`, since it needs no quoting at all. wait_text: Optional text to wait for after launch (e.g…

List all visible windows on the desktop. Optionally filter by title substring. Returns window titles, positions, sizes, and process names. Args: title_filter: Optional substring to filter window titles (case-insensitive).

Minimize a window by its title (substring match). Returns action result and a screenshot. Args: title: Window title substring to find and minimize.

Click the mouse at coordinates or current position. Returns action result and a screenshot of current screen state. Args: x: X screen coordinate. If not provided, clicks at current position. y: Y screen coordinate. If not provided, clicks at current position. button: Mouse button - 'left', 'right', or 'middle'. clicks: Number of clicks (2 for double-click).

Double-click at coordinates or current position. Returns action result and a screenshot of current screen state. Args: x: X screen coordinate. y: Y screen coordinate.

Drag from one point to another. Returns action result and a screenshot of current screen state. Args: start_x: Starting X coordinate. start_y: Starting Y coordinate. end_x: Ending X coordinate. end_y: Ending Y coordinate. button: Mouse button to hold ('left' or 'right'). duration: Duration of the drag in seconds.

Move the mouse cursor to screen coordinates. Args: x: X screen coordinate. y: Y screen coordinate.

Scroll the mouse wheel. Returns action result and a screenshot of current screen state. Args: amount: Scroll amount. Positive = up, negative = down. x: Optional X coordinate to scroll at. y: Optional Y coordinate to scroll at.

Report what changed on screen since the last observation. Prefer this over `screenshot` for tracking state. It maintains a running model of on-screen text and rescans only the regions that actually moved, so it costs a fraction of a full screen read and returns tens of tokens instead of thousands. The first call on an unfamiliar screen reads everything. If the screen has been seen before it is recognised and reused instead, after spot-checking that it really is the same screen. Args: force_full: Rescan the entire screen instead of only what changed. Use if you suspect th…

Report how much perception work has been avoided. Covers incremental scanning, remembered screens, and predicted actions.

Press a key or key combination. Returns action result and a screenshot of current screen state. Supports single keys and combos like 'Enter', 'Ctrl+C', 'Alt+Tab', 'Ctrl+Shift+S'. Args: key: Key name or combo (e.g., 'Enter', 'Ctrl+A', 'Alt+F4').

Read on-screen text from the incremental model. Unlike read_screen_text this does not re-OCR the display; it refreshes only the regions that changed and then answers from the model. Args: query: Optional substring filter. limit: Maximum elements to return.

Read ALL visible text on the screen using OCR. Returns every detected text element with its position, in absolute screen coordinates that are ready to click. Args: region_left: Optional left bound to restrict OCR area. region_top: Optional top bound to restrict OCR area. region_width: Optional width to restrict OCR area. region_height: Optional height to restrict OCR area. monitor: Monitor index.

Remember the current screen so future visits skip reading it. Useful after arriving at a screen you expect to return to, such as a settings page or a dialog. Remembered screens persist across sessions and are always spot-checked before being reused.

Find text on screen using OCR and right-click on it (context menu). Returns action result and a screenshot of current screen state. Args: text: The visible text to find and right-click. exact: Require exact text match. timeout: Maximum seconds to wait for the text to appear. poll_interval: Seconds between each retry. monitor: Monitor index.

Take a screenshot of the screen or a region. Returns the image as native MCP image content. Optionally saves to a file path. Args: save_path: Optional file path to save the screenshot. region_left: Left coordinate for a sub-region capture. region_top: Top coordinate for a sub-region capture. region_width: Width of the sub-region. region_height: Height of the sub-region. monitor: Monitor index (0 = all monitors, 1 = primary, etc.).

Take a screenshot of a specific window (by title). Useful for capturing just one application without the rest of the desktop. Returns the image as native MCP image content. Args: title: Window title substring to capture. save_path: Optional file path to save the screenshot.

Set text to the system clipboard. Useful for pasting data into applications. Args: text: The text to copy to the clipboard.

Type text using the keyboard (character by character). Returns action result and a screenshot of current screen state. Args: text: The text to type. delay: Delay in seconds between each character.

Wait for the screen to visually change. Takes a baseline screenshot, then polls until the screen looks different (or timeout). Useful after triggering an action to wait for the UI to update. Args: timeout: Maximum seconds to wait for a change. poll_interval: Seconds between polls.

Wait for specific text to appear on screen. Polls via OCR until found or timeout. Args: text: The text to wait for. exact: Require exact match. timeout: Maximum seconds to wait. Defaults to the server's --timeout. poll_interval: Seconds between polls. monitor: Monitor index.

Wait for text to disappear from screen. Polls via OCR until gone or timeout. Useful for waiting for loading spinners, progress dialogs, etc. to finish. Args: text: The text to wait to disappear. exact: Require exact match. timeout: Maximum seconds to wait. Defaults to the server's --timeout. poll_interval: Seconds between polls. monitor: Monitor index.

Wait for a specified amount of time, then take a screenshot. Use sparingly — prefer wait_for_text or wait_for_text_gone instead. Capped at 30 seconds maximum. Args: seconds: Number of seconds to wait (max 30).