Playwriter

Control your real Chrome browser with AI agents, CLI commands, or any MCP client. Full Playwright API access through your existing browser with all your logins, extensions, and cookies already there. Other browser automation tools spawn a fresh Chrome instance. That means no login state, no extensions, instant bot detection, and double the memory usage. Playwriter connects to your running browser instead. Click the extension icon on any tab to start automating it. HOW IT WORKS Click the Playwriter extension icon on any tab you want to control The icon turns green when connected Your AI agent, CLI, or MCP client can now control that tab through the full Playwright API The extension creates a local WebSocket relay on localhost:19988. Your automation tools connect to this relay and send Playwright commands. The extension forwards them to Chrome via the DevTools Protocol. Everything runs locally on your machine. WHAT YOU CAN DO Browser automation with full Playwright API. Navigate pages, click buttons, fill forms, take screenshots, intercept network requests, set breakpoints, live-edit page code, record videos, and more. All through a single execute tool that accepts any Playwright code. Accessibility snapshots give your agent a text representation of every interactive element on the page. 5-20KB instead of 100KB+ for a screenshot. Faster, cheaper, and the agent can parse them without vision capabilities. Visual labels overlay Vimium-style colored labels on interactive elements for agents that need spatial context. Yellow for links, orange for buttons, coral for inputs, pink for checkboxes. Session isolation lets multiple agents work in the same browser without interference. Each session has its own sandbox with persistent state. Network interception captures API calls behind JavaScript-rendered pages. Agents can reverse-engineer APIs, inspect schemas, and replay requests with different parameters. JavaScript debugger sets breakpoints, steps through code, inspects variables at runtime. Full Chrome DevTools Protocol access, not a limited subset. Live code editor views and edits page scripts and CSS in memory without reloading. Toggle debug flags, patch code, search across all loaded scripts. Screen recording captures the browser tab as MP4 video at 30-60fps using chrome.tabCapture. Recordings survive page navigation. Built-in ghost cursor overlay for creating demo videos. Cloud browsers run stealth Chromium instances in the cloud with automatic CAPTCHA solving Headless mode launches a headless Chrome automatically for CI, servers, and fully autonomous automation without a visible browser. Direct CDP connection bypasses the extension entirely to connect to Chrome instances with remote debugging enabled, Docker containers, or cloud browser providers via WebSocket. Remote access controls Chrome on another machine over the internet using secure tunnels. No VPN, no firewall rules, no port forwarding needed. USE CASES Web development: your coding agent verifies its own work in a real browser, checks rendered pages, takes screenshots, inspects CSS, sets breakpoints, and fixes issues in one loop. QA testing: run tests against your real Chrome profile with real login state. No test accounts or mock authentication needed. Task automation: automate repetitive browser tasks that need your logged-in session. Bulk fill forms, export data from dashboards, download files from authenticated sites. Scraping JavaScript-heavy sites: Instagram, Twitter, Facebook, and other SPAs return empty HTML to curl and fetch. Playwriter renders them in a real browser with your session cookies. API reverse engineering: capture network requests behind client-side rendering. Discover undocumented APIs, inspect their schemas, replay calls with different parameters. Remote support: control a user's browser remotely to fix issues. The user starts a tunnel, shares the URL, and you can see exactly what they see. Screen recording and demos: record browser automation as polished MP4 videos with cursor overlay. Speed up idle sections automatically for demo videos. WHY PLAYWRITER INSTEAD OF OTHER TOOLS vs Playwright MCP: Playwright MCP spawns a new Chrome with no logins, no extensions, and gets flagged by bot detectors. Playwriter uses your existing Chrome. vs BrowserMCP and other browser MCPs: other tools expose 12+ dedicated tools with high context window overhead. Playwriter uses one execute tool with the full Playwright API. Less context usage, more capabilities. vs Chrome CDP mode: Chrome's remote-debugging-port flag requires restarting Chrome and shows persistent "controlled by automated software" banners that interrupt your workflow. Playwriter runs silently with no confirmation dialogs. vs Claude Browser Extension: Playwriter works with any MCP client, not just Claude. Uses text snapshots (5-20KB) instead of screenshots (100KB+). Full CDP access including debugger, network interception, and live code editing. COLLABORATION Because the agent works in your browser, you can collaborate in real time. See everything the agent does. Solve captchas when it gets stuck. Click through consent walls. Disable the extension temporarily to bypass bot detection, then re-enable to continue automation. SECURITY AND PRIVACY Local only: the WebSocket server binds to localhost. Nothing leaves your machine. No remote servers, no accounts, no telemetry. Origin validation: only the Playwriter extension origin is accepted. Browsers cannot spoof the Origin header, so malicious websites cannot connect. Explicit consent: only tabs where you clicked the extension icon are controlled. No background access to other tabs. Visible automation: Chrome shows an automation banner on controlled tabs so you always know when a tab is being automated. No data collection: no analytics, no tracking, no telemetry. All browser control happens locally through Chrome DevTools Protocol. SUPPORTED MCP CLIENTS Works with Claude Desktop, Cur…