Browser Automation — Claude Code MCP server: browser control and setup
This MCP server lets an agent navigate websites, interact with page elements, fill forms, and capture screenshots. Use it for browser tasks where you want to see or resume the session.
claude mcp add --env AGENT_BROWSER_SESSION_NAME=<AGENT_BROWSER_SESSION_NAME> --env AGENT_BROWSER_SHOW_WINDOW=<AGENT_BROWSER_SHOW_WINDOW> --env MCP_WORKSPACE_PATH=<MCP_WORKSPACE_PATH> --transport stdio browser-automation -- npx -y @mindstone/mcp-server-browser-automationWhat Browser Automation does
An agent can open URLs, inspect page text and titles, interact with form fields, switch tabs, take screenshots, and save a page as a PDF. It can target elements through accessibility references or CSS selectors. Browser sessions can persist for later use.
The server needs Node.js 20+, npm, and the agent-browser CLI on PATH; it can also launch the CLI through npx. No API key or credentials are needed to connect. The browser window is visible by default, and MCP_WORKSPACE_PATH sets the directory used for PDFs and uploaded files. Without agent-browser on PATH or an available npx, the server cannot control a browser.
What it helps with
- open a page and retrieve its title
- fill in a website form
- capture the current page
- save a page as a PDF in the workspace
- switch to another browser tab
Example prompts for the agent
- Open the sign-in page and wait while I log in.
- Fill in the form on the current page with these details, then show me a screenshot before submitting.
- Save the current page as a PDF in the workspace.
How to install Browser Automation
Platform CLI
claude mcp add --env AGENT_BROWSER_SESSION_NAME=<AGENT_BROWSER_SESSION_NAME> --env AGENT_BROWSER_SHOW_WINDOW=<AGENT_BROWSER_SHOW_WINDOW> --env MCP_WORKSPACE_PATH=<MCP_WORKSPACE_PATH> --transport stdio browser-automation -- npx -y @mindstone/mcp-server-browser-automationReplace <…> with your own values
Platform CLI
codex mcp add browser-automation --env AGENT_BROWSER_SESSION_NAME=<AGENT_BROWSER_SESSION_NAME> --env AGENT_BROWSER_SHOW_WINDOW=<AGENT_BROWSER_SHOW_WINDOW> --env MCP_WORKSPACE_PATH=<MCP_WORKSPACE_PATH> -- npx -y @mindstone/mcp-server-browser-automationReplace <…> with your own values
Config file
{
"mcpServers": {
"browser-automation": {
"command": "npx",
"args": [
"-y",
"@mindstone/mcp-server-browser-automation"
],
"env": {
"AGENT_BROWSER_SESSION_NAME": "<AGENT_BROWSER_SESSION_NAME>",
"AGENT_BROWSER_SHOW_WINDOW": "<AGENT_BROWSER_SHOW_WINDOW>",
"MCP_WORKSPACE_PATH": "<MCP_WORKSPACE_PATH>"
}
}
}
}Replace <…> with your own values
Config file
{
"servers": {
"browser-automation": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@mindstone/mcp-server-browser-automation"
],
"env": {
"AGENT_BROWSER_SESSION_NAME": "<AGENT_BROWSER_SESSION_NAME>",
"AGENT_BROWSER_SHOW_WINDOW": "<AGENT_BROWSER_SHOW_WINDOW>",
"MCP_WORKSPACE_PATH": "<MCP_WORKSPACE_PATH>"
}
}
}
}Replace <…> with your own values
Platform CLI
gemini mcp add -e AGENT_BROWSER_SESSION_NAME=<AGENT_BROWSER_SESSION_NAME> -e AGENT_BROWSER_SHOW_WINDOW=<AGENT_BROWSER_SHOW_WINDOW> -e MCP_WORKSPACE_PATH=<MCP_WORKSPACE_PATH> browser-automation npx -y @mindstone/mcp-server-browser-automationReplace <…> with your own values
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about Browser Automation
How do I install Browser Automation in Claude Code?
Run claude mcp add mcp-server-browser-automation -- npx -y @mindstone/mcp-server-browser-automation. This registers the server through npx.
What does Browser Automation do?
It lets an agent control a browser: navigate pages, interact with page elements, manage tabs, and capture screenshots or PDFs.
What does Browser Automation need to run?
It requires Node.js 20+, npm, and either agent-browser on PATH or an available npx to launch the CLI. No API keys are needed.
