Extended Playwright MCP Server
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Extended Playwright MCP Server extract the table data from the page"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Extended Playwright MCP Server
This MCP server extends the official @playwright/mcp server by implementing 22 additional tools to facilitate web automation, direct uploads, element interactions, and advanced scraping workflows for AI agents. All standard Playwright tools remain 1:1 available and fully functional.
Custom Tools (Features)
1. download_and_convert_images
Description: Downloads an image from a given URL, converts it to PNG format, and saves it locally.
Parameters:
url(String, required): The URL of the image to download.outputDir(String, optional): The target directory (defaults to a directory nameddownloadsinside the current workspace).
2. browser_get_html
Description: Extracts the raw HTML markup of a specific element (by CSS selector or Snapshot ID) or the entire page.
Parameters:
target(String, optional): CSS selector or Snapshot ID. If omitted, the HTML of the entire page is returned.
3. browser_scroll_to
Description: Scrolls the page or a specific element to coordinates or preset positions.
Parameters:
target(String, optional): CSS selector or Snapshot ID of the element to scroll into view.x(Number, optional): Horizontal scroll coordinate.y(Number, optional): Vertical scroll coordinate.position(Enum["top", "bottom", "intoView"], optional): Scroll position preset.
4. browser_extract_table
Description: Finds and extracts HTML table data (
<table>or grid rows) and returns it as a structured JSON array.Parameters:
target(String, required): CSS selector or Snapshot ID of the table or grid element.
5. browser_wait_for_network_idle
Description: Blocks execution until all network requests on the page are completed (no network requests active for at least 500ms). Perfect for modern Single Page Applications (SPAs).
Parameters:
timeout(Number, optional): Maximum timeout in milliseconds (defaults to 10000ms).
6. browser_get_attributes
Description: Returns all HTML attributes of an element as a key-value object (e.g. class, style, data-* attributes).
Parameters:
target(String, required): CSS selector or Snapshot ID of the target element.
7. browser_handle_download
Description: Clicks an element, intercepts the triggered browser download event, and saves the file to the workspace.
Parameters:
target(String, required): CSS selector of the download button or link.filename(String, optional): Custom filename to save the file as locally.
8. browser_clear_input
Description: Deterministically clears the text content of any input field or textarea.
Parameters:
target(String, required): CSS selector or Snapshot ID of the input field.
9. browser_manage_storage
Description: Allows reading, writing, or clearing data from
localStorageorsessionStorage.Parameters:
action(Enum["get", "set", "clear"], required): The storage action to perform.storageType(Enum["local", "session"], required): The storage engine type.key(String, optional): The key to access or modify (required forgetandset).value(String, optional): The value to write (required forset).
10. browser_toggle_checkbox
Description: Checks or unchecks a checkbox or radio button element deterministically.
Parameters:
target(String, required): CSS selector or Snapshot ID of the target element.checked(Boolean, required): Whether the element should be checked (true) or unchecked (false).
11. browser_mock_route
Description: Intercepts and blocks network requests (e.g. blocking images/ads to save bandwidth/tokens) or fulfills them with mock HTTP responses.
Parameters:
urlPattern(String, required): Glob or regex pattern (e.g.**/*.{png,jpg}or**/api/v1/data).action(Enum["abort", "mock"], required): Action to perform.mockStatus(Number, optional): HTTP status code for mock responses.mockBody(String, optional): Response body for mock responses.
12. browser_upload_file_direct
Description: Directly uploads files to a file input element (
<input type="file">) without needing to trigger or wait for a browser file chooser modal.Parameters:
target(String, required): CSS selector or Snapshot ID of the file input element.paths(Array of Strings, required): List of absolute/relative file paths to assign.
13. browser_take_screenshot_element
Description: Captures a PNG/JPEG screenshot of a specific element and saves it to the workspace.
Parameters:
target(String, required): CSS selector or Snapshot ID of the target element.filename(String, optional): Custom filename to save the screenshot.type(Enum["png", "jpeg"], optional): Output image format (defaults topng).
14. browser_get_page_info
Description: Retrieves frame tree, URL, page title, user agent, tab count, and viewport information.
Parameters: None.
15. browser_set_geolocation
Description: Emulates a specific geographic location for testing region-restricted content.
Parameters:
latitude(Number, required): Latitude between -90 and 90.longitude(Number, required): Longitude between -180 and 180.accuracy(Number, optional): Geolocation accuracy value.
16. browser_emulate_media
Description: Emulates dark/light color scheme theme settings and media type queries (screen vs. print).
Parameters:
colorScheme(Enum["light", "dark", "no-preference"], optional): Emulated color scheme.media(Enum["screen", "print", "null"], optional): Emulated media type.
17. browser_focus_element
Description: Directs DOM focus onto the target element.
Parameters:
target(String, required): CSS selector or Snapshot ID of the element to focus.
18. browser_dispatch_event
Description: Dispatches a custom event (e.g. click, change, input) with optional custom initialization parameters on a target element.
Parameters:
target(String, required): CSS selector or Snapshot ID of the target element.type(String, required): Event type string (e.g.'change').eventInit(Object, optional): Custom event initialization parameters.
19. browser_get_element_state
Description: Checks current states of an element: whether it is visible, enabled, editable, or checked.
Parameters:
target(String, required): CSS selector or Snapshot ID of the element.
20. browser_key_down_up
Description: Separately fires keydown and keyup events for modifier controls or custom key holding.
Parameters:
key(String, required): Name of the key (e.g., 'Shift', 'Control', 'a', 'ArrowUp').action(Enum["down", "up"], required): Press down or release key.
21. browser_get_links
Description: Extracts and lists all internal and external link URLs from the current page.
Parameters: None.
22. browser_mouse_move_click
Description: Performs absolute mouse movement and click actions on coordinate pixels.
Parameters:
x(Number, required): X coordinate in pixels.y(Number, required): Y coordinate in pixels.click(Boolean, optional): Set to true to click at the coordinates.button(Enum["left", "right", "middle"], optional): Mouse click button.
Related MCP server: Kryfto
Integration
You can bind this local MCP server to your preferred MCP client (such as Claude Desktop or Cursor). Since it runs locally, you execute it using node and the path to the compiled JavaScript bundle.
1. Claude Desktop (Configuration)
Add the server to your Claude Desktop configuration file (e.g. at %APPDATA%\Claude\claude_desktop_config.json):
{
"mcpServers": {
"playwright": {
"command": "node",
"args": [
"C:/Users/danev/Downloads/extend-playwright-mcp/dist/index.js",
"--extension"
],
"env": {
"PLAYWRIGHT_MCP_EXTENSION_TOKEN": "PYefeTc_P7a5Ej8TmHCIshkWvtIPnlQ90H04ItZ66Aw"
}
}
}
}Note: Replace
C:/Users/danev/Downloads/extend-playwright-mcpwith your actual directory path if the project folder is moved. Ensure you use forward slashes (/) instead of backslashes (\) in the JSON paths.
2. Cursor (Configuration)
Add a new MCP server in Cursor:
Name:
playwrightType:
commandCommand:
node "C:/Users/danev/Downloads/extend-playwright-mcp/dist/index.js" --extensionEnvironment Variables:
Name:
PLAYWRIGHT_MCP_EXTENSION_TOKENValue:
PYefeTc_P7a5Ej8TmHCIshkWvtIPnlQ90H04ItZ66Aw
Development and Compilation
If you modify the TypeScript code, compile the project again using:
# Install dependencies
npm install
# Compile the project (outputs to the dist/ folder)
npm run buildAvailable Tools
46 toolsbrowser_clear_inputADestructive
Clears the text in an input or textarea element.
Inputs: target (required).
browser_clickBDestructive
Perform click on a web page
Inputs: target (required), button, element, modifiers, +1 more.
browser_closeBDestructive
Close the page
Inputs: no named parameters.
browser_console_messagesBRead-only
Returns all console messages
Inputs: level (required), all, filename.
browser_dispatch_eventBDestructive
Dispatches a custom event with optional custom initialization parameters on a target element.
Inputs: type (required), target (required), eventInit.
browser_dragADestructive
Perform drag and drop between two elements
Inputs: endTarget (required), startTarget (required), endElement, startElement.
browser_dropADestructive
Drop files or MIME-typed data onto an element, as if dragged from outside the page. At least one of "paths" or "data" must be provided.
Inputs: target (required), data, paths, element.
browser_emulate_mediaADestructive
Emulates light/dark color scheme or media type queries (screen vs. print).
Inputs: media, colorScheme.
browser_evaluateBDestructive
Evaluate JavaScript expression on page or element
Inputs: function (required), target, element, filename.
browser_extract_tableARead-only
Extracts tabular HTML table data into a structured JSON array.
Inputs: target (required).
browser_file_uploadBDestructive
Upload one or multiple files
Inputs: paths.
browser_fill_formBDestructive
Fill multiple form fields
Inputs: fields (required).
browser_findARead-only
Search the accessibility snapshot of the current page for text or a regular expression. Returns matching snapshot nodes with a few lines of surrounding context (like search snippets), each shown under its path from the root of the tree, which is cheaper than capturing the whole snapshot when you only need to locate an element and its ref.
Inputs: text, regex.
browser_focus_elementCDestructive
Focuses the target element.
Inputs: target (required).
browser_get_attributesARead-only
Returns all attributes of an HTML element as a key-value dictionary.
Inputs: target (required).
browser_get_element_stateARead-only
Checks the state of an element: visible, enabled, editable, checked, or focused.
Inputs: target (required).
browser_get_htmlARead-only
Returns the raw HTML of a target element or the entire page if no target is specified.
Inputs: target.
browser_get_linksARead-only
Extracts all links from the current page.
Inputs: no named parameters.
browser_get_page_infoARead-only
Retrieves detailed page metadata including URL, title, frames, tabs count, viewport, and agent information.
Inputs: no named parameters.
browser_handle_dialogCDestructive
Handle a dialog
Inputs: accept (required), promptText.
browser_handle_downloadADestructive
Clicks an element and intercepts the resulting file download event, saving the file to the workspace.
Inputs: target (required), filename.
browser_hoverADestructive
Hover over element on page
Inputs: target (required), element.
browser_key_down_upADestructive
Fires a separate keydown or keyup event for custom key combinations or modifier actions.
Inputs: key (required), action (required).
browser_manage_storageADestructive
Inspects, sets, or clears values in local/session storage.
Inputs: action (required), storageType (required), key, value.
browser_mock_routeADestructive
Intercepts and mocks/blocks network requests matching a URL pattern (e.g. blocking images/ads, or mocking responses).
Inputs: action (required), urlPattern (required), mockBody, mockStatus.
browser_mouse_move_clickCDestructive
Controls fine mouse movement to target coordinates and clicks on coordinates.
Inputs: x (required), y (required), click, button.
browser_network_requestARead-only
Returns full details (headers and body) of a single network request, or a single part if part is set. Use the number from browser_network_requests.
Inputs: index (required), part, filename.
browser_network_requestsARead-only
Returns a numbered list of network requests since loading the page. Use browser_network_request with the number to get full details.
Inputs: static (required), filter, filename.
browser_press_keyBDestructive
Press a key on the keyboard
Inputs: key (required).
browser_resizeBDestructive
Resize the browser window
Inputs: width (required), height (required).
browser_run_code_unsafeADestructive
Run a Playwright code snippet. Unsafe: executes arbitrary JavaScript in the Playwright server process and is RCE-equivalent.
Inputs: code, filename.
browser_scroll_toADestructive
Scrolls the page or a specific element into view or to custom coordinates.
Inputs: x, y, target, position.
browser_select_optionBDestructive
Select an option in a dropdown
Inputs: target (required), values (required), element.
browser_set_geolocationBDestructive
Emulates a specific geographic location for the browser context.
Inputs: latitude (required), longitude (required), accuracy.
browser_snapshotBRead-only
Capture accessibility snapshot of the current page, this is better than screenshot
Inputs: boxes, depth, target, filename.
browser_tabsADestructive
List, create, close, or select a browser tab.
Inputs: action (required), url, index.
browser_take_screenshotARead-only
Take a screenshot of the current page. You can't perform actions based on the screenshot, use browser_snapshot for actions.
Inputs: type (required), scale (required), target, element, +2 more.
browser_take_screenshot_elementARead-only
Takes a screenshot of a specific element and saves it to the workspace.
Inputs: target (required), type, filename.
browser_toggle_checkboxADestructive
Checks or unchecks a checkbox/radio button element deterministically.
Inputs: target (required), checked (required).
browser_typeBDestructive
Type text into editable element
Inputs: text (required), target (required), slowly, submit, +1 more.
browser_upload_file_directADestructive
Directly sets input files on a file input element bypassing the file-chooser modal requirement.
Inputs: paths (required), target (required).
browser_wait_forARead-only
Wait for text to appear or disappear or a specified time to pass
Inputs: text, time, textGone.
browser_wait_for_network_idleARead-only
Waits until there are no active network connections (network idle state) for at least 500ms.
Inputs: timeout.
download_and_convert_imagesADestructive
Downloads an image from the given URL and converts it to PNG format. Saves output in target directory and returns the absolute local path.
Inputs: url (required), outputDir.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 46 tools
Most tools have distinct purposes, but a few pairs overlap in function (e.g., browser_file_upload vs browser_upload_file_direct, browser_drag vs browser_drop) and could confuse an agent. Overall, descriptions help clarify the boundaries between similar actions.
The vast majority of tools follow the browser_<action> pattern, but there are notable exceptions like `download_and_convert_images` (no browser_ prefix) and noun-style names such as browser_tabs, browser_network_requests, and browser_console_messages. These deviations are minor but break full consistency.
With 46 tools, the server is heavily over-scoped for a browser automation toolkit. Many fine-grained actions like browser_key_down_up, browser_mouse_move_click, and browser_dispatch_event could be consolidated or covered by a generic evaluate tool. While the domain is broad, this count exceeds what is typically considered well-scoped.
The tool surface provides comprehensive coverage of browser automation: navigation, interaction, extraction, screenshots, network interception, storage management, tabs, dialogs, downloads, and geolocation. It also includes browser_run_code_unsafe to fill any remaining gaps, making the surface effectively complete for its stated purpose.
Maintenance
Related MCP Connectors
Web scraping, browser automation, screenshots and AI data extraction for AI agents via MCP.
Playwright browser automation — navigate, click, fill forms, extract, screenshot — 17 agent tools.
Control a real browser from AI agents: navigate, click, type, screenshot, read the page.
Screenshot, diff, audit and sitemap-capture any web page — 5 MCP tools for AI agents.
Related MCP Servers
AlicenseNot gradedqualityDmaintenanceEnables comprehensive browser automation and web interaction through Playwright with 50+ specialized functions for navigation, form filling, data extraction, and Chrome DevTools Protocol support. Designed specifically for AI agents to perform complex web workflows including scraping, testing, and automated browsing tasks.15 npm1Apache 2.0- FlicenseNot gradedqualityDmaintenanceProvides 42+ MCP tools for browser automation, web scraping, and search, enabling AI agents like Claude and Cursor to browse, extract data, and run research agents on the live web.9-
- FlicenseNot gradedqualityDmaintenanceExposes Playwright browser automation as MCP tools, enabling AI assistants to control a real browser tab-by-tab for form filling, navigation, and more, while preserving the user's active session.-
- FlicenseNot gradedqualityDmaintenanceEnables AI agents to browse the web, interact with pages, fill forms, and extract data using Playwright, with 30 tools across 7 categories.2-