Browser

You can use the MCP Browser Debugger to inspect and debug frontend pages end-to-end. It lets you open a visible browser, log in to a site, navigate pages, inspect the DOM, run JavaScript in the page context, capture network activity and console logs, and take screenshots—all through a programmable MCP server you control from your client.

How to use

To get started, first ensure the browser is open. Then perform a visible browser session, log in if needed, and proceed with inspection and debugging tasks such as viewing the DOM, running scripts, or monitoring network activity.

How to install

Install prerequisites and set up the MCP Browser Debugger locally.

1. Install dependencies and build the TypeScript project.

npm install
npm run build

Configuration and usage notes

Configure your MCP client to connect to the browser debugger as a local stdio server. The following configuration uses Node to run the built browser debugger from a local path.

{
  "mcpServers": {
    "browser_debugger": {
      "command": "node",
      "args": ["c:\\Users\\Emmanuel\\Desktop\\mcp\\mcp-browser\\dist\\index.js"]
    }
  }
}

Security and maintenance

• The browser runs in a mode that is suitable for debugging, with session data kept in memory during a session. - Sessions, cookies, and localStorage are cleared when the browser is closed per user instruction. - Logs and network activity accumulate between navigations; you can clear them when needed.

Notes and best practices

• Always verify the browser status before opening a new instance to avoid losing an active login session. - Build the project after any code changes before restarting the MCP client variant. - Use the provided command flow to ensure you maintain a single login session throughout debugging.

Available tools

open_browser

Opens the browser in visible mode to show actions in real time.

close_browser

Closes the browser and ends the session, clearing cookies and storage.

login

Performs automated login to a specified URL and keeps the session active for inspection.

navigate

Navigates to a given URL with timeout and wait conditions.

get_dom

Extracts the complete HTML of the current DOM.

get_element

Inspects a specific element and returns computed styles and attributes.

query_selector_all

Finds multiple elements using a CSS selector and returns matches.

evaluate_xpath

Evaluates an XPath expression against the current document.

get_page_source

Extracts the full page source including scripts and styles.

execute_js

Executes arbitrary JavaScript in the page context and returns the result.

get_console_logs

Retrieves console logs accumulated during the session.

get_network_activity

Monitors network activity and returns HTTP requests and responses.

screenshot

Captures a screenshot of the full page or a specific element.

click_element

Clicks on a specified element and optionally waits for navigation.

type_text

Types text into an input field, with optional clear before typing.

get_local_storage

Reads localStorage entries from the page.

get_cookies

Lists cookies stored for the current page.

get_page_info

Returns general information about the current page.

get_page_info

Returns general information about the current page.

get_page_info

Returns general information about the current page.

evaluate_xpath

Evaluates an XPath expression against the current document.