Potencia tu agente
El comando propio del publicador. No es un endpoint de PlugMCP.
Host a copy with our keys and logs.
npx -y [email protected]Claude Code no tiene enlace de instalación para un comando local. Copia el comando de abajo y ejecútalo en una terminal.
claude mcp add agentic-playwright-mcp -- npx -y agentic-playwright-mcp@1.2.2Ejecútalo en una terminal.
{ "mcpServers": { "agentic-playwright-mcp": { "command": "npx", "args": [ "-y", "[email protected]" ] } }}claude_desktop_config.json
¿URL remota o un comando en tu máquina? En qué se diferencian las dos instalaciones
Ejecutar en PlugMCP
A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.
AbrirDel publicador
agentic-playwright-mcp lets multiple agents control one Chrome instance without sending
commands to the wrong tab. Each browser tool accepts a targetId. Agents can also use a
groupId to list and control only their own tabs.
This package is an independent fork of Playwright MCP. It is not a Microsoft package.
Fork of official Playwright MCP with multi-agent tab isolation via targetId
Use this command as the server command in your MCP client:
npx --yes agentic-playwright-mcp
For clients that use JSON configuration, add:
{
"mcpServers": {
"agentic-playwright-mcp": {
"command": "npx",
"args": ["--yes", "agentic-playwright-mcp"]
}
}
}
The command starts an MCP server over standard input and output. It waits for requests from an MCP client, so running it directly does not open a prompt or web page.
Chrome starts when the client calls the first browser tool. The server uses one headed
Chrome instance and stores its profile in ~/.agentic-playwright-mcp.
To restart Chrome if it exits, add --keep-alive to the arguments:
"args": ["--yes", "agentic-playwright-mcp", "--keep-alive"]
To connect to an existing browser, start Chrome with remote debugging and add
--cdp-endpoint:
"args": [
"--yes",
"agentic-playwright-mcp",
"--cdp-endpoint",
"http://localhost:9222"
]
Install the package globally if you do not want the MCP client to run it through npx:
npm install --global agentic-playwright-mcp
apmcp --help
The package installs these commands:
agentic-playwright-mcp and apmcp start the MCP server.agentic-playwright-mcp-daemon and apmcp-daemon manage the Chrome daemon.browser_tab_group with action: "create". Save the returned groupId.browser_tabs with action: "new" and the groupId. Save the returned
targetId.targetId and groupId to later browser calls.Selecting a tab does not raise Chrome. If the window is minimized or behind another app,
it stays there. Tools keep using targetId, so the tab does not have to be in front.
Cookies, local storage, and sign-in state remain shared because all groups use the same
Chrome profile. A person can complete an interactive sign-in, then the agent can continue
on the same targetId.
The package includes a companion extension and loads it through CDP when Chrome starts. When Chrome allows the extension, group listings include native groups that a person made in the tab strip. If browser policy blocks the extension, MCP group isolation still works.
browser_run_code_unsafe runs JavaScript with access to the selected Playwright page.
Only let trusted clients call this tool.
The Chrome DevTools Protocol endpoint listens on localhost:9223 by default. Do not
expose this port to other machines.
git clone https://github.com/TechyAditya/agentic-playwright-mcp.git
cd agentic-playwright-mcp
npm ci
npm run build
npm test
npm run test:e2e
Use Playwright MCP when one agent owns the browser. Use this package when agents must share a signed-in browser without sharing tab control.
This project includes work by Patrick Menlove and code from OpenClaw. See the license file for attribution and license terms.
Fork of official Playwright MCP with multi-agent tab isolation via targetId
{ "mcpServers": { "agentic-playwright-mcp": { "command": "npx", "args": [ "-y", "[email protected]" ] } }}Herramientas
Aún no hay lista de herramientas. Los servidores remotos la reciben tras la primera comprobación de estado.
Preguntas frecuentes
Copia en esta página la configuración stdio para Cursor, Claude Desktop, Claude Code o VS Code. Usa la instalación del publicador, no un proxy de PlugMCP.
Este registro no indica ninguna variable de entorno obligatoria.
Agentic Playwright se ejecuta localmente por stdio.
Última comprobación: 2026-10-27. Fork of official Playwright MCP with multi-agent tab isolation via targetId