CodingBaby-Browser-MCP is a powerful tool that allows AI agents like Claude 3.7 Sonnet in Cursor to control your Chrome browser for automated tasks. With this tool, your AI assistant can:
- Fill out web forms automatically
- Perform automated testing of websites
- Navigate through complex web applications
- Publish content to websites
- Take screenshots of web pages
- Extract data from websites
- Execute any browser-based workflow
This tool requires two components to work properly:
- The MCP Tool Server - Connects your AI assistant to your browser
- The Chrome Extension - Allows the MCP tool to control your browser
- Open Cursor and click on Settings
- Select MCP
- Click Add new global MCP server
- Add the following JSON configuration (be careful with commas if you have other MCP tools):
{
"mcpServers": {
"CodingBaby-Browser-MCP": {
"command": "npx",
"args": ["@sydneyassistent/codingbaby-browser-mcp"]
}
// Your other MCP tools may be here
}
}
- Visit the Chrome Web Store at: CodingBaby Extension
- Click "Add to Chrome"
- Follow the prompts to complete installation
For the best experience, make sure to use Claude 3.7 Sonnet as your AI model, as it provides superior instruction-following capabilities and visual interaction support.
- Go to Cursor → Settings → MCP
- Click the "Refresh" button to reload MCP tools
- If the MCP status indicator turns green, your connection is working properly
If the MCP status indicator is red in Cursor:
- Click the "Refresh" button to restart the MCP connection
- Wait a few seconds for the connection to re-establish
If you see an error about port 9876 being in use:
- Click "Refresh" in Cursor's MCP settings
- The tool will automatically attempt to resolve the conflict
To get the latest version of the CodingBaby-Browser-MCP:
- Go to Cursor → Settings → MCP
- Click the "Refresh" button
- The tool will automatically pull and install the latest version
Once everything is installed, ask Claude 3.7 in Cursor to perform a simple browser task, such as:
"Use the CodingBaby-Browser-MCP to open Google's homepage"
If successful, you'll see your Chrome browser automatically open Google's homepage.
The following browser control tools are available:
-
navigate
: Go to any URL -
click
: Click at specific coordinates on the page -
type
: Enter text into forms -
press_key
: Simulate keyboard actions -
scroll
: Scroll in any direction -
area_screenshot
: Capture specific areas of the screen -
wait
: Pause for a specified duration -
batch
: Execute multiple operations in sequence -
set_viewport
: Change browser window size -
tab_new
: Open new browser tabs -
tab_list
: List all open tabs -
tab_select
: Switch between tabs -
tab_close
: Close one tab -
close
: Close all tabs