[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"pack-detail-mcp-browser-automation-stack-fr":3,"seo:pack:mcp-browser-automation-stack:fr":95},{"code":4,"message":5,"data":6},200,"操作成功",{"pack":7},{"slug":8,"icon":9,"tone":10,"status":11,"status_label":12,"title":13,"description":14,"items":15,"install_cmd":94},"mcp-browser-automation-stack","🌐","#06B6D4","new","Nouveau · cette semaine","Stack d'Automatisation Navigateur MCP","Dix choix pour brancher un vrai navigateur dans Claude Code, Cursor ou Gemini CLI via MCP. Playwright MCP d'abord pour des snapshots déterministes, Chrome DevTools MCP pour le debug live, puis Chrome MCP \u002F Puppeteer MCP \u002F Browserbase MCP selon les tradeoffs. Proxies multi-navigateurs, fleet avec login partagé, fixes de popup et auditeur de logs ferment la marche. Dans l'ordre.",[16,28,36,43,50,57,64,73,80,87],{"id":17,"uuid":18,"slug":19,"title":20,"description":21,"author_name":22,"view_count":23,"vote_count":24,"lang_type":25,"type":26,"type_label":27},152,"290f10c7-59c4-49aa-bfdd-56e8becfa222","playwright-mcp-server-browser-automation-ai-290f10c7","Playwright MCP Server — Browser Automation for AI","Official Playwright MCP server for browser automation. Lets Claude Code, Cursor, and other AI tools navigate pages, fill forms, click buttons, take screenshots, and run end-to-end tests via natural la","Microsoft AI",169,0,"en","mcp","MCP",{"id":29,"uuid":30,"slug":31,"title":32,"description":33,"author_name":34,"view_count":35,"vote_count":24,"lang_type":25,"type":26,"type_label":27},3059,"7f3a756c-8ec2-4b20-8ad6-99bfc5155af3","playwright-mcp-browser-automation-server","Playwright MCP — Browser Automation Server","Playwright MCP is an MCP server for browser automation via Playwright snapshots. Add via npx in Claude Code\u002FCodex to run deterministic actions.","MCP Hub",149,{"id":37,"uuid":38,"slug":39,"title":40,"description":41,"author_name":34,"view_count":42,"vote_count":24,"lang_type":25,"type":26,"type_label":27},477,"17f46368-d971-4227-ab65-f5f183e7dedc","chrome-devtools-mcp-browser-debugging-ai-agents-17f46368","Chrome DevTools MCP — Browser Debugging for AI Agents","Give your AI coding agent full access to Chrome DevTools for browser automation, debugging, and performance analysis. Works with Claude, Cursor, Copilot, and 15+ AI tools.",207,{"id":44,"uuid":45,"slug":46,"title":47,"description":48,"author_name":34,"view_count":49,"vote_count":24,"lang_type":25,"type":26,"type_label":27},3060,"e10ca754-dc37-41c7-b94a-fdad9e5de1bf","chrome-mcp-server-extension-based-browser-mcp","Chrome MCP Server — Extension-Based Browser MCP","mcp-chrome turns Chrome into an MCP server via an extension + bridge. Install mcp-chrome-bridge, load the extension, then connect to 127.0.0.1:12306\u002Fmcp.",61,{"id":51,"uuid":52,"slug":53,"title":54,"description":55,"author_name":34,"view_count":56,"vote_count":24,"lang_type":25,"type":26,"type_label":27},675,"efa97531-0468-4c2b-ab24-aff9b4090ac3","puppeteer-mcp-headless-chrome-server-ai-agents-efa97531","Puppeteer MCP — Headless Chrome Server for AI Agents","MCP server wrapping Google Puppeteer for headless Chrome automation. Navigate pages, screenshot, generate PDFs, scrape SPAs, and test web apps through AI agent commands. 2,000+ stars.",121,{"id":58,"uuid":59,"slug":60,"title":61,"description":62,"author_name":34,"view_count":63,"vote_count":24,"lang_type":25,"type":26,"type_label":27},3192,"33a22c18-fe2c-4870-9a56-5f86ac0991cb","browserbase-mcp-cloud-browser-automation-tools","Browserbase MCP — Cloud Browser Automation Tools","Browserbase MCP server exposes automation tools (navigate, act, observe, extract) backed by Browserbase + Stagehand, letting agents operate real web pages.",66,{"id":65,"uuid":66,"slug":67,"title":68,"description":69,"author_name":70,"view_count":71,"vote_count":24,"lang_type":72,"type":26,"type_label":27},2610,"deaee53e-ad94-4671-a766-5c6aeb5f186c","multi-browser-mcp-proxies-arc-browser-chrome-beta-variants-deaee53e","Multi-Browser MCP Proxies — Arc Browser & Chrome Beta Variants","Companion to 'Chrome MCP Background Proxy' for running parallel, isolated MCP fleets against Arc Browser and Chrome Beta on top of the same cdp-proxy.mjs. Arc-specific proxy auto-discovers the WebSocket path from \u002Fjson\u002Fversion (Arc doesn't write a DevToolsActivePort file in the standard location); Chrome Beta proxy points at Beta's own DevToolsActivePort. Lets you run mcp__chrome__*, mcp__beta__*, and mcp__arc__* side-by-side with independent client state and no cross-talk.","henuwangkai",192,"",{"id":74,"uuid":75,"slug":76,"title":77,"description":78,"author_name":70,"view_count":79,"vote_count":24,"lang_type":72,"type":26,"type_label":27},2611,"20bc3ffd-1d7a-41d1-86d0-b668e8500cee","chrome-fleet-multi-agent-browser-pool-shared-login-state-20bc3ffd","Chrome Fleet — Multi-Agent Browser Pool with Shared Login State","Multi-agent control plane for chrome-devtools-mcp. Two modes: (1) shared main Chrome — N CDP proxies on 9401\u002F9402\u002F9403... all multiplexing onto one logged-in Chrome :9222 so every agent inherits your real cookies\u002Fextensions, with focus protection and ID isolation handled by cdp-proxy.mjs; (2) isolated agent Chromes — dedicated Chrome instance per agent on :930N with its own user-data-dir for multi-account \u002F persona-isolation testing. Includes a status tool to inspect the running fleet.",186,{"id":81,"uuid":82,"slug":83,"title":84,"description":85,"author_name":70,"view_count":37,"vote_count":86,"lang_type":25,"type":26,"type_label":27},498,"29944683-2033-4e30-9449-6d18dffa60f5","chrome-mcp-background-proxy-fix-popups-focus-stealing-multi-29944683","Chrome MCP Background Proxy — Fix Popups, Focus Stealing & Multi-Agent Conflicts","Persistent CDP proxy + entry script that lets chrome-devtools-mcp run against your real, logged-in Chrome without the Chrome 146+ consent popup spamming on every connection, without focus stealing (Target.activateTarget \u002F Page.bringToFront are intercepted, createTarget is forced to background), and without request-ID \u002F event collisions when multiple Claude Code windows or sub-agents share one Chrome. Includes the proxy core (cdp-proxy.mjs v3), entry script, safe cleanup, pre-flight healthcheck, and a launchd-style self-healing watchdog with Feishu alerts.",1,{"id":88,"uuid":89,"slug":90,"title":91,"description":92,"author_name":34,"view_count":93,"vote_count":24,"lang_type":25,"type":26,"type_label":27},3819,"80bc9d31-f1e5-5528-8e4b-bc93fe9b1d88","browsertoolsmcp-audit-browser-logs-via-mcp","BrowserToolsMCP — Audit Browser Logs via MCP","BrowserToolsMCP streams browser logs to MCP clients: install via `npx`, run a local server, and audit network\u002Fconsole events from your IDE (7.2k★).",57,"tokrepo install pack\u002Fmcp-browser-automation-stack",{"pageType":96,"pageKey":8,"locale":25,"title":97,"metaDescription":98,"h1":99,"tldr":100,"bodyMarkdown":101,"faq":102,"schema":118,"internalLinks":124,"citations":137,"wordCount":150,"generatedAt":151},"pack","MCP Browser Automation Stack — 10 Servers to Wire a Real Browser into Claude \u002F Cursor \u002F Gemini","Playwright MCP, Chrome DevTools MCP, Chrome MCP, Puppeteer MCP, Browserbase MCP, multi-browser proxies, shared-login fleet, popup-focus fix, and BrowserToolsMCP — an opinionated order for plugging a real browser into your AI agent. Cookies, iframes, debug, CI.","MCP Browser Automation Stack — Plug a Real Browser into Your Agent","Ten MCP servers in deliberate install order. Playwright MCP for deterministic snapshot-driven actions, Chrome DevTools MCP for live network and console debugging, Chrome MCP \u002F Puppeteer MCP \u002F Browserbase MCP for different runtime tradeoffs, multi-browser proxies for iframe and Arc \u002F Beta variants, Chrome Fleet for cookie sharing across agents, a popup-focus fix, and BrowserToolsMCP for log auditing. End state: your agent drives a browser the way a senior QA does.","## What this pack actually solves\n\nThe MCP browser servers all promise the same thing — let your AI assistant click, type, and read a webpage. In practice they make different tradeoffs around determinism, cookie scope, iframe support, debug visibility, and CI-friendliness. This pack picks ten servers that together cover those tradeoffs explicitly, and orders them so each install unlocks the next layer.\n\nThe target user is a developer wiring a browser into **Claude Code**, **Cursor**, or **Gemini CLI** for end-to-end testing, scraping behind a login, agent-driven dogfooding, or live UI debugging. By the end of the install order your agent can pick the right MCP for a given task instead of brute-forcing one server into every shape.\n\n## Install in this order\n\n1. **Playwright MCP Server (official, Microsoft)** — start here. Snapshot-based actions are deterministic, the API surface is small, and it works against Chromium, Firefox, and WebKit. The accessibility-tree snapshot is what makes agents reliable: they read structured node IDs instead of guessing CSS selectors. According to the Playwright docs the snapshot format is the recommended interface for AI clients.\n2. **Playwright MCP — Browser Automation Server (community npx)** — same engine, lighter setup via `npx`. Use this when you want to bolt onto an existing Playwright install without the official package's bundling. Good fit for Codex \u002F Cursor power users.\n3. **Chrome DevTools MCP** — the second MCP you install, not the first. It hangs off DevTools Protocol and exposes network, console, performance traces, and DOM inspection. Pair it with Playwright MCP: Playwright drives the page, Chrome DevTools MCP tells you *why* it broke.\n4. **Chrome MCP Server (extension-based)** — different model. Runs as a Chrome extension instead of a separate browser process, which means it reuses your **real logged-in profile** — cookies, extensions, saved passwords. Pick this when the site is gated by SSO or anti-bot and a fresh Playwright context gets challenged.\n5. **Puppeteer MCP — Headless Chrome Server** — the lean choice for CI. No extra cross-browser layer, smaller install, faster cold start. Use it for pure Chromium scraping pipelines where you don't need Firefox \u002F WebKit fallbacks.\n6. **Browserbase MCP — Cloud Browser** — the escape hatch when local Chromium hits IP blocks, stealth requirements, or scale ceilings. Cloud-hosted browser pool with stealth profiles; the MCP server exposes the same tool shape so agent code is portable.\n7. **Multi-Browser MCP Proxies (Arc \u002F Chrome Beta variants)** — the iframe-aware layer. Stock Chrome MCP can't always reach into Arc's split panes or Beta's variant DOM trees; this proxy multiplexes between Arc \u002F Beta \u002F Chrome Stable so the agent picks the right window. Critical when the site renders core UI inside an iframe (most embedded checkouts, Stripe, Cloudflare challenges).\n8. **Chrome Fleet — Multi-Agent Browser Pool with Shared Login State** — once you have multiple agents, login fatigue dominates. Fleet keeps a warm pool of authenticated browsers and hands each agent a checked-out instance. Cookie scope is per-fleet, not per-agent, which is the design tradeoff: faster onboarding, weaker isolation. Use it when your agents trust each other.\n9. **Chrome MCP Background Proxy — Popup \u002F Focus Fix** — the *bug-fix* MCP. Stock Chrome MCP loses focus when a popup or OAuth window opens and the agent's next click lands on the wrong tab. This proxy intercepts focus events and reroutes them deterministically. Install this **before** you wire up any login-required flow, not after you debug a flaky run.\n10. **BrowserToolsMCP — Audit Browser Logs via MCP** — closes the loop. After the agent finishes a session, BrowserToolsMCP lets it (or you) replay console errors, network failures, and page lifecycle events. This is the difference between \"the test passed\" and \"the test passed and emitted zero errors.\"\n\n## How they fit together\n\n```\n          Playwright MCP (deterministic, cross-browser)\n                │\n                ├── Chrome DevTools MCP (why it broke)\n                │\n          Chrome MCP (real profile)  ←  Multi-Browser Proxy (Arc \u002F Beta)\n                │\n          Puppeteer MCP (CI lean)      Chrome Fleet (shared login)\n                │\n          Browserbase MCP (cloud, stealth)\n                │\n          Chrome MCP Background Proxy (popup focus fix)\n                │\n          BrowserToolsMCP (log audit)\n```\n\nThe spine is **Playwright MCP + Chrome DevTools MCP + BrowserToolsMCP** — that trio handles 80% of agent browser work for a CI-style developer. Chrome MCP \u002F Chrome Fleet are the **real-profile branch** when your site is locked behind SSO. Browserbase MCP is the **escape hatch**.\n\n## Cookies, sessions, iframes — pick the right scope\n\n- **Per-context cookies (Playwright MCP, Puppeteer MCP, Browserbase MCP)** — each session starts clean. Best for tests. Worst for sites with bot detection that flags first-visit fingerprints.\n- **Real-profile cookies (Chrome MCP extension)** — agent reuses your logged-in Chrome profile. Best for SSO-gated dashboards. Worst for parallel agents (they fight over the same cookie jar).\n- **Fleet-shared cookies (Chrome Fleet)** — agents share a warmed-up pool. Best for many agents on one tenant. Worst for tenant isolation; you would not run two customers' agents on the same fleet.\n- **Iframe handling** — Playwright MCP exposes frame snapshots natively; Chrome DevTools MCP can navigate cross-origin frames via CDP; Chrome MCP extension is limited to same-origin frames unless you ship a content script. If your target embeds Stripe \u002F Cloudflare Turnstile \u002F Auth0, plan for Playwright MCP plus the Multi-Browser proxy.\n\n## Debug-to-CI path\n\n1. **Develop locally** with Playwright MCP + Chrome DevTools MCP. Run headed, watch the agent drive the page, read DevTools panel in real time.\n2. **Switch to BrowserToolsMCP** to catch silent errors you missed — console warnings, failed XHRs, hydration mismatches.\n3. **Move to CI** by swapping Playwright MCP for Puppeteer MCP (lighter) and running headless. If anti-bot blocks the CI IP, swap again to Browserbase MCP and keep the same tool calls.\n4. **Production agent fleet** — Chrome Fleet for shared login, Background Proxy enabled to handle OAuth popups, BrowserToolsMCP scraping logs for the post-mortem dashboard.\n\n## Common pitfalls\n\n- **Installing Chrome MCP extension and Playwright MCP at the same time** — they race for the same DevTools port. Pick one as the primary; use the other for fallback only.\n- **Forgetting `npx playwright install`** — the official Playwright MCP requires the browser binaries. The MCP server fails opaquely if they're missing.\n- **Cookie scope confusion** — agents that worked in dev (real profile) silently break in CI (fresh context) because the dashboard requires SSO. Decide cookie scope on day one.\n- **iframe blindness** — Stagehand \u002F Browser-Use style agents that work on simple pages crash on embedded checkouts. If your target uses iframes, default to Playwright MCP plus the multi-browser proxy.\n- **OAuth popup focus theft** — install the Background Proxy *before* the first login flow, not after the third flaky run.\n",[103,106,109,112,115],{"q":104,"a":105},"Which MCP do I install first if I only have time for one?","Playwright MCP (the official Microsoft server). Snapshot-based actions give your agent the most deterministic interface — it reads structured accessibility-tree IDs instead of guessing CSS selectors, and the same API works across Chromium, Firefox, and WebKit. Every other MCP in this pack is an optimization on top. If you skip Playwright MCP and start with Chrome MCP or Puppeteer MCP, you trade determinism for either real-profile cookies or smaller install footprint — that's the right call only after you know which constraint matters most.",{"q":107,"a":108},"What's the difference between Playwright MCP and Chrome DevTools MCP — don't they both drive Chrome?","They drive Chrome at different layers. Playwright MCP sits on the Playwright automation library and exposes action verbs like 'navigate', 'click', 'fill'. Chrome DevTools MCP sits directly on the Chrome DevTools Protocol and exposes inspection verbs like 'get network requests', 'read console messages', 'profile performance'. You use both: Playwright MCP to make things happen, Chrome DevTools MCP to figure out why they didn't. Running only Chrome DevTools MCP forces the agent to assemble actions from low-level CDP primitives, which is fragile.",{"q":110,"a":111},"How do I handle sites that need a logged-in session?","Three options ordered by cost. (1) Playwright MCP with a saved storage state — script the login once, save cookies to a JSON file, replay on every session. Cleanest for tests. (2) Chrome MCP extension — reuses your real Chrome profile cookies, no scripting. Good for one-developer dashboards, breaks with parallel agents. (3) Chrome Fleet — warmed-up pool of authenticated browsers shared across agents. Good for multi-agent setups in a single tenant. Mix only if you have to: race conditions on cookies are the most painful bug in this stack.",{"q":113,"a":114},"Will any of these work in CI \u002F Docker without a display?","Yes, but pick the right one. Playwright MCP and Puppeteer MCP both run headless in containers — Puppeteer MCP has a smaller image footprint and faster cold start. Chrome MCP extension does not work in CI because it requires a real Chrome profile and extension load. Browserbase MCP is the cloud escape hatch when you want to run from CI but the target site blocks your CI IP range. Chrome DevTools MCP and BrowserToolsMCP both work headless because they hang off the DevTools Protocol, not a window.",{"q":116,"a":117},"What about iframes — most checkout flows embed Stripe or Cloudflare challenges?","Iframes are where most agent stacks silently fail. Playwright MCP exposes frame snapshots natively, so the agent can navigate into cross-origin iframes by ID. Chrome DevTools MCP can attach to any frame via CDP target IDs. Chrome MCP extension is limited to same-origin frames unless you ship a content script alongside. If your flow goes through Stripe, Cloudflare Turnstile, or Auth0 Universal Login, default to Playwright MCP, install the Multi-Browser Proxy from this pack, and test the iframe path before anything else — fixing it later means rewriting tool calls.",{"@context":119,"@type":120,"name":121,"description":122,"numberOfItems":123,"inLanguage":25},"https:\u002F\u002Fschema.org","ItemList","MCP Browser Automation Stack","Ten MCP servers in deliberate install order to plug a real browser into Claude Code, Cursor, or Gemini CLI — covering Playwright, Chrome DevTools, Chrome extension, Puppeteer, Browserbase, multi-browser proxies, fleet, popup fix, and log audit.",10,[125,129,133],{"url":126,"anchor":127,"reason":128},"\u002Fen\u002Fai-tools-for\u002Fbrowser-automation","Browser automation tools on TokRepo","Adjacent category covering the non-MCP agents (Browser-Use, Skyvern, Stagehand) that pair with these MCP servers",{"url":130,"anchor":131,"reason":132},"\u002Fen\u002Ftopics\u002Fmcp-server-stack","MCP Server Stack pack","Sibling pack covering the broader MCP server menu beyond browser",{"url":134,"anchor":135,"reason":136},"\u002Fen\u002Fmcp\u002Fchrome","Chrome MCP family on TokRepo","Drill-down into the Chrome-specific MCP servers featured in this pack",[138,142,146],{"claim":139,"source_name":140,"source_url":141},"Playwright supports Chromium, Firefox, and WebKit through one API","Playwright documentation","https:\u002F\u002Fplaywright.dev\u002Fdocs\u002Fintro",{"claim":143,"source_name":144,"source_url":145},"Model Context Protocol specifies how AI clients connect to tool servers","Model Context Protocol","https:\u002F\u002Fmodelcontextprotocol.io\u002F",{"claim":147,"source_name":148,"source_url":149},"Chrome DevTools Protocol exposes network, console, and performance APIs","Chrome DevTools Protocol","https:\u002F\u002Fchromedevtools.github.io\u002Fdevtools-protocol\u002F",1109,"2026-05-22T00:00:00Z"]