Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| BROWSER_INSTANCE_TTL | No | Session TTL in seconds (default: 30 minutes) | 1800 |
| BROWSER_EXECUTE_TIMEOUT | No | Command execution timeout in seconds | 30 |
| BROWSER_CLEANUP_INTERVAL | No | Cleanup interval in seconds | 60 |
| BROWSER_MAXIMUM_INSTANCES | No | Maximum number of concurrent browser instances | 10 |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| create_chrome_instance | Create a new Chrome browser instance and return session_id (UUID) |
| close_instance | Close a specific Chrome browser instance by session_id, will delete the instance and all related data |
| get_instance_info | Get detailed information about a specific browser instance |
| check_browser_health | Check the health status of a browser session and provide recovery suggestions |
| get_browser_status | Get current browser status, including all instances and their status,only admin can use this tool |
| close_all_instances | Close all browser instances and delete all related data, only admin can use this tool |
| set_browser_config | Set advanced browser configuration for a specific instance. If you want to change the viewport, you can use the set_browser_config tool to change the viewport_width and viewport_height. |
| get_browser_config | Get advanced browser configuration for a specific instance |
| navigate_to | Navigate to URL in current tab or new tab (does not auto-switch to new tab). |
| navigate_back | Go back in current tab's history. |
| navigate_forward | Navigate forward in browser history |
| refresh_page | Refresh the current page |
| get_page_state | Get current page state and interactive elements; frames mode is unstable and may not extract all details. |
| get_tabs_info | Get information about all open browser tabs |
| close_tab | Close a specific tab, if the tab is the last one, it will goto about:blank |
| switch_tab | Switch to a specific browser tab and return the tab info object |
| click_element_by_xpath | Click an interactive element by XPath with confirmation |
| click_element | Click an interactive element by index with confirmation |
| input_text | Input text into a form field |
| set_element_value | Set value of an input or select element directly, supports frames |
| get_element_info | Get detailed information about a DOM element (supports index or xpath query), can get the value of the element |
| send_keys | Send keyboard keys to the browser |
| scroll_page | Scroll page up or down |
| upload_file | Upload file to file input element |
| generate_pdf | Generate PDF from current page, URL, or HTML content |
| take_screenshot | Take screenshot with structured result, it will take screenshot of the target element if target is not None, otherwise it will take screenshot of the full page, otherwise it will take screenshot of the current viewport Screenshot page or element (by target selector, supports CSS/XPath, e.g. 'button', '//button', 'css=button', 'xpath=//button'). See: https://playwright.dev/python/docs/locators#locate-by-css-or-xpath |
| download_file | Download any file from URL to the temp directory. Use fetch+blob+a.download, if failed, fallback to goto+expect_download, maximum compatibility. |
| download_image | Download a image from URL, it will download the image to the temp directory, and return the file path, it will open a new tab if the image is not from the same origin as the current page |
| set_cookie | Set a cookie in the browser |
| get_cookies | Get cookies from the browser |
| wait | Wait for specified number of seconds with progress |
| search_bing | Search Bing for a query with progress tracking |
| extract_content | Extract content from page based on query, including all frames/iframes |
| get_dropdown_options | Get options from a dropdown/select element, it will return the options of the element, and the id and name of the element |
| browser_tips | LLM-oriented best practices and practical tips for robust, step-by-step use of browser MCP tools. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
| web_testing | Generate a prompt for web testing scenarios |
| data_extraction | Generate a prompt for extracting data from web pages |
| form_filling | Generate a conversation for automated form filling |
| automation_troubleshooting | Generate a prompt for troubleshooting browser automation issues |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
| get_browser_status_resource | Get current browser status as a resource, only admin can use this tool |
| get_instances_resource | Get detailed information about all browser instances, only admin can use this tool |
| get_help_resource | Get concise help: all tools and resources, grouped and described. |