Let your agents control your own Chrome, via CLI or MCP. Your logins, extensions, cookies — already there.

Other browser MCPs spawn a fresh Chrome — no logins, no extensions, instantly flagged by bot detectors, double the memory. Playwriter connects to your running browser instead. One Chrome extension, full Playwright API, everything you're already logged into.

	Playwright MCP	Playwriter
Browser	Spawns new Chrome	Uses your Chrome
Extensions	None	Your existing ones
Login state	Fresh	Already logged in
Bot detection	Always detected	Can bypass (disconnect extension)
Collaboration	Separate window	Same browser as user

Installation

1. Install Extension from Chrome Web Store

2. Click extension icon on a tab → turns green when connected

3. Install the CLI and start automating the browser:

   npm i -g playwriter
   playwriter -s 1 -e "await page.goto('https://example.com')"

4. Install the skill so your agent knows how to use Playwriter:

   npx -y skills add remorses/playwriter

Quick Start

playwriter session new  # creates stateful sandbox, outputs session id (e.g. 1)
playwriter -s 1 -e "await page.goto('https://example.com')"
playwriter -s 1 -e "console.log(await snapshot({ page }))"
playwriter -s 1 -e "await page.locator('aria-ref=e5').click()"

CLI Usage

Each session has isolated state. Browser tabs are shared across sessions.

# Session management
playwriter session new              # creates stateful sandbox, outputs id (e.g. 1)
playwriter session list             # show sessions + state keys
playwriter session reset <id>       # fix connection issues

# Execute (always use -s)
playwriter -s 1 -e "await page.goto('https://example.com')"
playwriter -s 1 -e "await page.click('button')"
playwriter -s 1 -e "console.log(await page.title())"

Create your own page to avoid interference from other agents:

playwriter -s 1 -e "state.myPage = await context.newPage(); await state.myPage.goto('https://example.com')"

Multiline:

playwriter -s 1 -e $'
const title = await page.title();
console.log({ title, url: page.url() });
'

Examples

Variables in scope: page, context, state (persists between calls), require, and Node.js globals.

Persist data in state:

playwriter -e "state.users = await page.$$eval('.user', els => els.map(e => e.textContent))"
playwriter -e "console.log(state.users)"

Intercept network requests:

playwriter -e "state.requests = []; page.on('response', r => { if (r.url().includes('/api/')) state.requests.push(r.url()) })"
playwriter -e "await Promise.all([page.waitForResponse(r => r.url().includes('/api/')), page.click('button')])"
playwriter -e "console.log(state.requests)"

Set breakpoints and debug:

playwriter -e "state.cdp = await getCDPSession({ page }); state.dbg = createDebugger({ cdp: state.cdp }); await state.dbg.enable()"
playwriter -e "state.scripts = await state.dbg.listScripts({ search: 'app' }); console.log(state.scripts.map(s => s.url))"
playwriter -e "await state.dbg.setBreakpoint({ file: state.scripts[0].url, line: 42 })"

Live edit page code:

playwriter -e "state.cdp = await getCDPSession({ page }); state.editor = createEditor({ cdp: state.cdp }); await state.editor.enable()"
playwriter -e "await state.editor.edit({ url: 'https://example.com/app.js', oldString: 'const DEBUG = false', newString: 'const DEBUG = true' })"

Screenshot with labels:

playwriter -e "await screenshotWithAccessibilityLabels({ page })"

MCP Setup

Using the CLI with the skill (step 4 above) is the recommended approach. For direct MCP server configuration, see MCP.md.

Visual Labels

Vimium-style labels for AI agents to identify elements:

await screenshotWithAccessibilityLabels({ page })
// Returns screenshot + accessibility snapshot with aria-ref selectors
await page.locator('aria-ref=e5').click()

Color-coded: yellow=links, orange=buttons, coral=inputs, pink=checkboxes, peach=sliders, salmon=menus, amber=tabs.

Comparison

vs BrowserMCP

	BrowserMCP	Playwriter
Tools	12+ dedicated tools	1 execute tool
API	Limited actions	Full Playwright
Context usage	High (tool schemas)	Low
LLM knowledge	Must learn tools	Already knows Playwright

vs Antigravity (Jetski)

	Jetski	Playwriter
Tools	17+ tools	1 tool
Subagent	Spawns for each browser task	Direct execution
Latency	High (agent overhead)	Low

vs Claude Browser Extension

	Claude Extension	Playwriter
Agent support	Claude only	Any MCP client
Windows WSL	No	Yes
Context method	Screenshots (100KB+)	A11y snapshots (5-20KB)
Playwright API	No	Full
Debugger/breakpoints	No	Yes
Live code editing	No	Yes
Network interception	Limited	Full
Raw CDP access	No	Yes

Architecture

+---------------------+     +-------------------+     +-----------------+
|   BROWSER           |     |   LOCALHOST       |     |   MCP CLIENT    |
|                     |     |                   |     |                 |
|  +---------------+  |     | WebSocket Server  |     |  +-----------+  |
|  |   Extension   |<--------->  :19988         |     |  | AI Agent  |  |
|  +-------+-------+  | WS  |                   |     |  +-----------+  |
|          |          |     |  /extension       |     |        |        |
|    chrome.debugger  |     |       |           |     |        v        |
|          v          |     |       v           |     |  +-----------+  |
|  +---------------+  |     |  /cdp/:id <--------------> |  execute  |  |
|  | Tab 1 (green) |  |     +-------------------+  WS |  +-----------+  |
|  | Tab 2 (green) |  |                               |        |        |
|  | Tab 3 (gray)  |  |     Tab 3 not controlled      |  Playwright API |
+---------------------+     (no extension click)      +-----------------+

Remote Access

Control Chrome on a remote machine over the internet using traforo tunnels:

On host:

npx -y traforo -p 19988 -t my-machine -- npx -y playwriter serve --token <secret>

From remote:

export PLAYWRITER_HOST=https://my-machine-tunnel.traforo.dev
export PLAYWRITER_TOKEN=<secret>
playwriter -s 1 -e "await page.goto('https://example.com')"

Also works on a LAN without traforo (PLAYWRITER_HOST=192.168.1.10). Full guide with use cases (remote Mac mini, user support, multi-machine control): docs/remote-access.md

Security

• Local only: WebSocket server on localhost:19988
• Origin validation: Only our extension IDs allowed (browsers can't spoof Origin)
• Explicit consent: Only tabs where you clicked the extension icon
• Visible automation: Chrome shows automation banner on controlled tabs
• No remote access: Malicious websites cannot connect

Playwright API

Connect programmatically (without CLI):

import { chromium } from 'playwright-core'
import { startPlayWriterCDPRelayServer, getCdpUrl } from 'playwriter'

const server = await startPlayWriterCDPRelayServer()
const browser = await chromium.connectOverCDP(getCdpUrl())
const page = browser.contexts()[0].pages()[0]

await page.goto('https://example.com')
await page.screenshot({ path: 'screenshot.png' })
// Don't call browser.close() - it closes the user's Chrome
server.close()

Or connect to a running server:

npx -y playwriter serve --host 127.0.0.1

const browser = await chromium.connectOverCDP('http://127.0.0.1:19988')

Troubleshooting

View relay server logs to debug issues:

playwriter logfile  # prints the log file path
# typically: ~/.playwriter/relay-server.log

The relay log contains extension, MCP and WebSocket server logs. A separate CDP JSONL log is also created alongside it (see playwriter logfile). Both are recreated on each server start.

Example: summarize CDP traffic counts by direction + method:

jq -r '.direction + "\t" + (.message.method // "response")' ~/.playwriter/cdp.jsonl | uniq -c

Support

If Playwriter is useful to you, consider sponsoring the project.

Known Issues

• If all pages return about:blank, restart Chrome (Chrome bug in chrome.debugger API)
• Browser may switch to light mode on connect (Playwright issue)