# SeleniumBase MCP [Health: Active]

**Category:** 📂 Browser Automation  
**Repository:** https://github.com/seleniumbase/SeleniumBase  
**GitHub Stars:** 13250  
**Views:** 2  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/seleniumbase-mcp

## Description
Stealthy browser automation, testing, and web-scraping via CDP Mode.

## Claude Desktop Quick Installation
No generated config for this server: it installs through its own setup steps.

The project's own setup steps (from https://github.com/seleniumbase/SeleniumBase), copied verbatim. Run them in order; they are not an mcpServers config.

1. Install:

```sh
pip install seleniumbase
```

2. Install:

```sh
git clone https://github.com/seleniumbase/SeleniumBase.git
```

3. Run:

```sh
cd SeleniumBase/
```

4. Install:

```sh
pip install -e .
```

## Documentation

## What SeleniumBase MCP does

SeleniumBase MCP brings SeleniumBase’s browser automation capabilities into an MCP-based workflow. SeleniumBase is a Python framework for web scraping, browser automation, and end-to-end testing. Its CDP Mode provides a browser-control path designed to reduce bot-detection signals, and the framework includes options for working with CAPTCHA-protected pages.

The underlying SeleniumBase APIs can launch a browser, visit URLs, find elements, enter text, click controls, check page conditions, wait between actions, and close the browser. These capabilities make the server relevant to agents that need to perform multi-step web tasks rather than only retrieve static page content.

## How it works

Automation is built around SeleniumBase’s Python APIs. In CDP Mode, a Chrome browser is launched through SeleniumBase’s CDP-oriented interface, after which an automation flow can navigate between pages and interact with selectors. The framework also supports a pure CDP Mode interface through the `sb_cdp` package entry point.

A typical flow opens a browser, navigates to a target URL, checks whether expected elements are present, fills in fields, clicks controls, optionally handles a CAPTCHA, and quits the browser. SeleniumBase examples use direct Python execution for scripts and pytest for examples whose filenames begin or end with `test`.

## Tools and capabilities

- Launch and control a Chrome browser in CDP Mode.
- Navigate to web pages and perform multi-step interactions.
- Locate elements with selectors and text-based conditions.
- Type into fields and click page controls.
- Assert that elements exist or are not visible.
- Support web-scraping and end-to-end testing workflows.
- Use SeleniumBase’s CAPTCHA-solving capability in supported automation flows.
- Run browser checks with Python or pytest.

## Limitations and notes

CDP Mode is intended for browser automation where bot-detection resistance matters, but website behavior and CAPTCHA handling can vary by target. The SeleniumBase project contains broader testing features, reporting, screenshots, recorder functionality, and other APIs; an agent workflow should use only the capabilities exposed by the MCP server.

Browser automation remains dependent on the target site’s selectors, navigation behavior, authentication requirements, and anti-automation controls. Actions that modify accounts or submit forms should be reviewed before execution, especially when an agent is operating without human approval.

SeleniumBase is distributed under the MIT License. The project is Python-based, and its examples distinguish between pytest test files and scripts executed with the standard Python interpreter.

_Full upstream README: https://allmcps.com/mcp/seleniumbase-mcp/readme_