[Feature]: Extend browser_mouse_click_xy with button, clickCount, and delay options

[oleksii-tumanov]

🚀 Feature Request

The current browser_mouse_click_xy tool in the MCP browser mouse API is limited to basic left-clicks at coordinates.
This restricts its use in advanced automation scenarios, such as:

• Right-clicking for context menus on non-element areas (e.g., canvases or overlays).
• Double/triple-clicking for selecting words or triggering custom events.
• Simulating realistic user delays for slower interactions.

Example

Right click
await page.mouse.click(450, 300, { button: 'right'});

Double click
await page.mouse.click(450, 300, { button: 'left', clickCount: 2, delay: 100 });

Motivation

Aligning it with Playwright's full page.mouse.click(x, y, options) API would make low-level coordinate-based interactions more versatile, especially in vision-based tools where element refs aren't available (e.g., after browser_snapshot).

Without this, users must chain separate browser_mouse_down, browser_mouse_up, and browser_mouse_move calls, which is verbose and error-prone.

[Skn0tt]

Hi! There's a balance to keep here around detailedness of our MCP tools. Have you tried instructing your agent to use browser_run_code with these APIs, as an escape hatch?

[oleksii-tumanov]

Thanks for the feedback @Scn0tt.
I totally agree with keeping the MCP slim and understand the maintenance overhead.

In this specific case, my reasoning for the extension is that delay, double and right clicks covers basic flow and can be usefilul for 'human' interactions
Those operations are already supported by Playwright.
I'll update PR following suggestions.

However, if you feel this doesn't fit the MCP vision, it would be helpful to understand where the line between possible 'feature' and an 'escape hatch'. I'm a big fun of Playwright and want to make useful impact.