This topic primarily describes the atomic operations on browsers and usage methods of Browser MCP Server provided by Volcano Engine Function as a Service (veFaaS).

Feature description

veFaaS provides Browser-Use MCP Server to enable browser automation capabilities. In addition, veFaaS provides Browser MCP Server that performs various atomic operations on browsers. This allows you to perform fine-grained management and operations on browsers, web elements, keyboards, forms, tabs, screenshots, and PDF generation.

This topic primarily describes how to locally or remotely use a browser by using Browser MCP Server. The simple sample code is as follows:

The following table describes the parameters that you can add to the args field based on your business requirements.

Parameter	Description
--allowed-origins <origins>	The list of allowed origins for browser requests, which are separated by semicolons (;). By default, all origins are allowed.
--blocked-origins <origins>	The list of blocked origins for browser requests, which are separated by semicolons (;). The blocked origin list takes precedence over the allowed origin list. If the allowlist is not used, all the requests will be allowed except those sent from the sources in the blocklist.
--block-service-workers	Blocks service workers.
--browser <browser>	The browser or Chrome channel that you want to use. Valid values: chrome: Google Chrome browser firefox: Mozilla Firefox browser webkit: WebKit engine, which is used by Safari msedge: Microsoft Edge browser
--caps <caps>	The list of additional features to be enabled. The vision and pdf capabilities are supported and can be configured concurrently with commas (,) as separators. --caps vision,pdf: enables both visual and PDF features. --caps vision: only enables the visual feature. --caps pdf: only enables the PDF feature.
--cdp-endpoint <endpoint>	The endpoint of Customer Data Platform (CDP) to be connected.
--config <path>	The path of the browser configuration file. If you do not specify this parameter, the browser configuration information will be saved in the default configuration file path based on the installed browser and its version.
--device <device>	The device to be simulated. Example: iPhone 15.
--executable-path <path>	The path of the browser executable file.
--headless	After you specify this parameter, the browser will run in headless mode. If you do not specify this parameter, the browser will run in headed mode by default.
--host <host>	The host to which the server is bound, which is localhost by default. To bind to all network interfaces, you can set the value to 0.0.0.0.
--ignore-https-errors	Ignores HTTPS errors.
--isolated	Store the browser configuration file in memory instead of the disk.
--image-responses <mode>	Specifies whether to send an image response to the client. Valid values: allow: sends an image response. This value is used by default. omit: does not send an image response.
--no-sandbox	Disables the sandbox.
--output-dir <path>	The directory path of the output file.
--port <port>	The port on which the Server-Sent Events (SSE) protocol listens.
--proxy-bypass <bypass>	The list of domain names that bypass the proxy. The domain names are separated by commas (,), such as ".com,chromium.org,.domain.com".
--proxy-server <proxy>	Specify a proxy server, such as "http://myproxy:3128" and "socks5://myproxy:8080".
--save-trace	Save the Playwright traces of sessions to the output file directory.
--storage-state <path>	The path at which the storage status file is saved for isolated sessions.
--user-agent <ua string>	Specify a user agent string.
--user-data-dir <path>	The directory path of user data. If you do not specify this parameter, a temporary directory will be created.
--viewport-size <size>	Specify the viewport size of the browser in pixels. Example: "1280, 720".

Usage methods

You can use Browser MCP Server through the stdio (standard input & output) method.

Invoke a remote browser

{
"mcpServers": {
"vefaas-browser": {
"command": "npx",
"args": [
"@faas-mcp/browser@latest"
"--cdp-endpoint ws://sd1lqecn6rmou3g0tmej0.apigateway-cn-beijing.volceapi.com/v0.1/browsers/6f55a5c3-f495-4ea5-863e-55765abf0839/devtools/browser/3fc922d4-a668-412b-bdc3-46baa5018193?faasInstanceName=vefaas-ydej8hci-sq34afy9ap-d1rotceg68p2mudjluo0-sandbox"
]
},
]
}

Invoke a local browser

{
"mcpServers": {
"vefaas-browser": {
"command": "npx",
"args": [
"@faas-mcp/browser@latest"
]
}
}
}

Supported tools

Interactive tools

Navigation tools

Resource-related tools

Utility tools

Tab-related tools

Testing tools

Visual pattern tool

Tool Name	Description	Parameter
browser_snapshot	Capture page structure for automation.	None.
browser_click	Click the web page element.	element: The element description ref: The element reference
browser_drag	Perform the drag-and-drop operation.	startElement: The source element description startRef: The source element reference endElement: The target element description endRef: The target element reference
browser_hover	Move the pointer over the element.	element: The element description ref: The element reference
browser_type	Enter text in the field.	element: The element description ref: The element reference text: Text to be entered submit: Submit the text after entering. slowly: Enter the text character by character.
browser_select_option	Select options from a drop-down menu.	element: The element description ref: The element reference values: The option to be selected
browser_press_key	Press a key on the keyboard.	key: The name or character of a key
browser_wait_for	Wait until the conditions are met.	time: The number of seconds to wait text: Text to appear textGone: Text to disappear
browser_file_upload	Upload the file.	paths: The path to which the file is uploaded
browser_handle_dialog	Manage the dialog boxes of the browser.	accept: Accept the dialog box. promptText: The prompt text

Tool Name	Description	Parameter
browser_navigate	Open the specified URL.	url: the target URL
browser_navigate_back	Return to the previous page.	None.
browser_navigate_forward	Go to the next page.	None.

Tool Name	Description	Parameter
browser_take_screenshot	Capture a page image.	raw: the uncompressed image Filename: the path at which the file is saved element: the element to be captured ref: the element reference
browser_pdf_save	Export the page as a PDF file.	Filename: the path at which the file is saved
browser_network_requests	List network activities.	None.
browser_console_messages	Display console output.	None.

Tool Name	Description	Parameter
browser_install	Configure the browser environment.	None.
browser_close	Exit the current page.	None.
browser_resize	Change the window size.	width: the window width height: the window height

Tool Name	Description	Parameter
browser_tab_list	Display the open tabs.	None.
browser_tab_new	Create a tab.	url: the initial URL
browser_tab_select	Switch to an active tab.	index: the tab position
browser_tab_close	Delete a tab.	index: the tab to be closed

Tool Name	Description	Parameter
browser_screen_capture	Obtain a visual snapshot.	None.
browser_screen_move_mouse	Position the cursor.	Element: the element description x: the horizontal position y: the vertical position
browser_screen_click	Click at the coordinates.	Element: the element description x: the horizontal position y: the vertical position
browser_screen_drag	Drag between points.	element: the element description startX/startY: the start position endX/endY: the end position
browser_screen_type	Enter text at the cursor.	text: text to be entered Submit: Press Enter after entering.

Tool Name	Description	Parameter
browser_generate_playwright_test	Create an automated test.	name: the test identifier description: the test purpose Steps: the test steps