Skip to directory

About SeleniumBase MCP

SeleniumBase publishes three local MCP server variants for browser automation: Pure CDP Mode, WebDriver, and the broader SB API. The documented CDP server provides browser navigation, inspection, interaction, session, and page-output tools.

What you can use it for

  • Browser automation
  • Browser inspection and page extraction
  • Web testing and interaction
Connection and setup
Hosting
Self-hosted local process
Transport
stdio
Endpoint
Local process; no hosted endpoint documented
Authentication
No MCP authentication documented. Browser sessions use local browser resources; proxy or target-site authentication is configured by the user outside this catalog.

Requirements

  • Python 3.10 or later.
  • uv package manager.
  • For the driver and SB variants, the project documents downloading ChromeDriver; the Pure CDP variant does not use WebDriver.

Setup steps

  1. Clone https://github.com/seleniumbase/seleniumbase-mcp and change to the repository directory.
  2. Run `uv sync` to create the environment and install the MCP project and its `seleniumbase[mcp]` dependency.
  3. For the Pure CDP server, configure the MCP client to run `uv --directory <absolute path to seleniumbase-mcp> run seleniumbase-cdp` over stdio.
  4. For the WebDriver or SB variants, use the documented `seleniumbase-driver` or `seleniumbase-sb` command instead; the latter two require `uv run seleniumbase get chromedriver`.
  5. Reconnect the client and inspect tools/list. The project documents `start_browser`, `open_url`, and `click_element` among the exposed tools.
Tools 8+

The repository documents three different server variants with different tool surfaces. This entry captures representative documented tools; the complete inventory differs by selected server.

  • start_browser

    Start a browser session with the selected launch options.

    Parameters: url, headless, use_chromium, browser_executable_path, incognito, guest, ad_block, proxy
    01
  • close_browser

    Close the browser session.

    02
  • open_url

    Navigate the active browser to a URL.

    03
  • get_page_info

    Return current browser status, URL, title, origin, user agent, and history.

    04
  • find_elements

    Find page elements by selector.

    05
  • click_element

    Click a selected page element.

    06
  • get_content

    Read page content in a documented output format.

    07
  • solve_captcha

    Attempt supported CAPTCHA challenge handling; support is limited and not guaranteed.

    08

This server’s available tools can vary by account, permissions, or configuration. The list shown is not its full runtime catalog.

Limitations and notes
  • Browser sessions are stateful and a server instance holds one global session.
  • The project includes CAPTCHA-solving and anti-bot behavior; sites? access rules still apply, and the project warns solving is not universal.
  • The tool list varies among the CDP, Driver, and SB server variants.
  • No MCP initialize, tools/list, or representative tool call was performed for this catalog entry.
Original sources
  • Official publisher MCP repository README

    Documents official ownership, three server variants, Python/uv requirements, exact install/client setup, stdio transport, representative tools, CAPTCHA limitations, and ChromeDriver requirements by variant.

  • Official product website and brand asset

    Official SeleniumBase website supplies the downloaded 128x128 PNG product mark, visually reviewed on light and dark contact sheets.

  • official tool documentation

    Official docs describe consolidated selector/action tools; exact tools exposed section is authoritative and the stored eight-name inventory is incomplete.

Documentation checked 2026-10-28T00:00:00Z. Connection has not been runtime tested.

Similar tools