# Browser Use MCP Guide: Give Your AI Agent a Real Browser > Give Claude, Cursor or any AI agent a real cloud browser with Browser Use MCP: setup, ready prompts, Ultrafast presets, replay links and judge-checked results. Page: https://flymy.ai/blog/browser-use-mcp-guide/ [Blog](https://flymy.ai/blog/) / Browser Use MCP guide Guide ยท Browser automation # Browser Use MCP: give your AI agent a real browser A practical guide to browser automation for AI agents: connect Browser Use MCP to Claude, Cursor or Codex, pick the right model, and use everything new in this release - Ultrafast presets, follow-ups while a task runs, read-only replay links and judge-checked results. By **the FlyMy.AI team** Published 11 November 2026 Updated 11 November 2026 15 min read ![Illustration: five browser windows fly forward in formation. The front one shows a web form filled in by an AI agent with a green check mark, another a replay timeline with a play button, another a stopwatch.](https://flymy.ai/blog/browser-use-mcp-guide/img/hero-1536.webp) In short - **Browser Use MCP gives an AI assistant a real cloud browser.** You describe the task in plain words; Browser Use's agent does it in Browser Use Cloud and returns the result, the status, the cost and any files. - **No Browser Use account or key on FlyMy.AI.** Usage costs Browser Use's own price with 0% markup. - **New in this release:** bu-ultrafast and bu-fast presets, follow-ups while a task runs, continuing a finished session, public read-only replay links, judge-checked results and thinking levels up to max. - **Measured on 11 November 2026:** bu-ultrafast read a page title in about 4 seconds for $0.0012, and filled in and submitted a seven-field form in 11 seconds for $0.0024. ## What is a browser agent? A browser agent is an AI model that operates a web browser to finish a task. It opens pages, reads them, clicks, types, scrolls and fills in forms until the job is done, then reports what it found. You describe the outcome in plain words - "list the plans on these three pricing pages", "fill in this form with test data" - instead of writing selectors and scripts that break when a page changes. [Browser Use](https://github.com/browser-use/browser-use) is an open-source agent built for this: it is MIT-licensed, runs on any LLM, and works locally or on Browser Use Cloud, the company's hosted service. **Browser Use MCP** is how an AI assistant reaches it. The Model Context Protocol (MCP) lets Claude, Cursor, Codex and other clients call tools on a server, and the [FlyMy.AI MCP](https://flymy.ai/mcp/) includes Browser Use among its 1,500+ tools. The assistant you already use can hand a task to a cloud browser, follow it, and use the result in the rest of its work. ## Why give your AI agent a real browser Search and fetch tools cover a lot, but much of the web is built for people rather than for APIs: pricing pages that render with JavaScript, search filters, multi-step forms and dashboards behind a login. A browser reaches whatever a person can reach. - **It works on pages, not endpoints.** The agent uses a site the way a visitor does, so it needs no API, no scraping rules and no selectors. - **It runs in the cloud.** Browser Use Cloud hosts the browser, so nothing runs on your laptop and the task keeps going while your assistant does other work. - **You can watch and step in.** The person who started a task gets a private live view and can take over for a login, a 2FA code or a CAPTCHA. - **Logins carry over.** A saved browser profile keeps a login for the next task. - **It leaves evidence.** Each task returns its output, status and cost. Recordings, screenshots and downloads can be saved as files, and now a read-only replay can be shared. ## What's new in Browser Use MCP This release moves FlyMy.AI's Browser Use tools to Browser Use Cloud v4, with gpt-6-luna as the default model. Each change is a parameter or an action your assistant can call, and the [platform log](https://flymy.ai/changelog/) lists every one of them. ### Ultrafast presets `bu-ultrafast` is Browser Use's fastest preset for short interactive flows; `bu-fast` reasons more per step. ### Follow-ups while a task runs `send_task` queues a new instruction for a running task, or with `interrupt: true` the task acts on it now. ### Continue a finished session A follow-up keeps the conversation, files and saved profile, even after the browser has closed. ### Keep the browser for 30 minutes `keep_alive: true` holds it open after the task, for manual testing or the next step. ### Public replay links `set_browser_share_link` turns a read-only replay of the task, its steps and screenshots on or off. ### Judge-checked results `judge: true` returns a verdict with its reasoning; the task is a success only when it passes. ### Thinking up to max `thinking_level` takes `xhigh` and `max` on models that offer them, such as gpt-6-luna. ### Works out of the box No Browser Use key: FlyMy.AI's shared account, at Browser Use's own price with 0% markup. 4 s to read a page title on bu-ultrafast 11 s to fill in and submit a seven-field form 0% markup on Browser Use usage Our own test sessions on 11 November 2026; times and costs as Browser Use Cloud reported them. ## How to set up Browser Use MCP in a minute You add one MCP server to your assistant, sign in and ask. There is no Browser Use account to create and no key to copy. ### 1. Add FlyMy.AI to your assistant The FlyMy.AI MCP lives at one URL: `https://mcp-agents.flymy.ai/mcp`. - **Claude Code:** run `claude mcp add --transport http flymyai https://mcp-agents.flymy.ai/mcp`, then `/mcp` to sign in. - **Codex:** run `codex mcp add flymy --url https://mcp-agents.flymy.ai/mcp`, then `codex mcp login flymy`. - **Claude and Claude Desktop:** Settings, Connectors, Add custom connector, then paste the URL. - **Cursor and other MCP clients:** add the same URL as a remote MCP server. ### 2. Sign in Sign in with your FlyMy.AI account when the client asks. Tasks run on FlyMy.AI's Browser Use account and are billed to your FlyMy.AI balance; with your own Browser Use plan, you can save your key on the connection instead. ### 3. Ask for a task in plain words Your assistant finds the Browser Use actions by itself. Start with something small: Try it ``` Open example.com in a browser on bu-ultrafast and tell me the page title. ``` ### 4. Follow, steer and stop The assistant waits with `wait_for_browser_task` and shows you the output and the cost. Then it can send a follow-up, make a replay link or stop the browser with `stop_browser_task`, which you want once you are done. Prefer to let the assistant do the setup? Paste this into any assistant: Setup request ``` Read https://flymy.ai/skill.md and https://flymy.ai/mcp/browser-use.md. Help me connect Browser Use MCP to this assistant. Then help me run this task: Open news.ycombinator.com in a real browser, collect the top 10 story titles and links, and show me the recording. ``` ## Pick the right model: bu-ultrafast, bu-fast or gpt-6-luna The model decides how fast the agent moves and how carefully it reads each page. New sessions use gpt-6-luna; the two Ultrafast presets are built for speed. - `bu-ultrafast` fastest **Short, interactive flows.** A form, a few clicks, a quick read. Our tests: a page title in 4 s, a seven-field form in 11 s. - `bu-fast` preset **Quick, with more reasoning.** Thinks more per step than bu-ultrafast, for flows with a few more decisions in them. No thinking_level, like bu-ultrafast. - `gpt-6-luna` default **Economical and capable.** Takes every thinking level from disabled to max, for research, comparisons and harder pages. Our test: a judged two-step read in 16 s. The Ultrafast presets take no `thinking_level` and stay fixed for the session, so choose them when you start the task; they run without the managed proxy unless you set `proxy_country_code`. On gpt-6-luna, `thinking_level` goes from `disabled` to `max`; use `xhigh` or `max` for pages that need careful reading, such as a long comparison. Try it ``` Compare three project-management tools in a browser with thinking_level max. ``` ## Steer a browser task while it runs Long tasks rarely go exactly to plan: halfway through, you want one more field, a different filter or a second page. Until this release you waited for the task to end and started again. Now `send_task` reaches a task while it is still running. How `send_task` reaches a task: queued for next, run at once with `interrupt: true`, or continued after the task, in the same browser while it is open. - **Queued (default).** The instruction goes into Browser Use's session queue and the task takes it up next. The response carries a `queued_message_id`, and `wait_for_browser_task` waits until the queued instruction has run. - **Interrupt.** With `interrupt: true` the task cancels its current work and runs your instruction now. - **After it finishes.** A follow-up continues the same session with its conversation, files and saved profile: in the same browser while it is still open, or in a fresh browser once it has closed. `keep_alive: true` keeps the browser open up to 30 minutes after the task. A running task takes only the instruction text. To change `model`, `thinking_level` or `judge`, send the follow-up after the task finishes; a session on an Ultrafast preset keeps its preset. Try it ``` Start a browser task that compares the pricing pages of Asana, Trello and ClickUp and lists each plan with its monthly price. While it is still running, tell it to also note which plans include a free trial. ``` ## Share a read-only replay of a browser session Sometimes you want to show how the agent got there: to a client, a teammate or a support thread. `set_browser_share_link` turns on a public, read-only link to the session's replay on Browser Use's page, with the task, the agent's steps and its screenshots, and no control of the browser. Turn it off when you are done; turning it on again restores the same link. ![Browser Use replay page of a FlyMy.AI test session on gpt-6-luna. The task reads: open flymy.ai/mcp/browser-use, read the page heading, scroll to the table of actions and return the first five action names. Below it: Worked for 16s, the agent's steps (open page and inspect heading, scroll to actions table) and its answer, the heading Browser Use MCP: automate browsers with no API key and a numbered list of five actions.](https://flymy.ai/blog/browser-use-mcp-guide/img/replay-steps.webp) The replay page of one of our own test sessions on 11 November 2026: the task, the agent's steps and its answer. We turned the link off after taking this screenshot. **Share only what you would show.** Anyone with the link sees what the session saw, including pages behind a login. The live view is different: it gives full control of a running browser and is meant only for the person who started the task. Never post it. Try it ``` Make a public read-only replay link of my last browser task. ``` When the review is over, ask: "Turn off the replay link of my last browser task." ## Check results with the judge A browser agent can report success and still miss part of the task. `judge: true` asks Browser Use's LLM judge to grade the finished task against its instruction. The verdict comes back as `judgement`, with `verdict`, `reasoning` and `failure_reason`, and `is_success` follows it: the task counts as a success only when the verdict passes. The judge adds a small model charge to the task. With the judge on, a task's success is the judge's verdict, not the agent's own report. In the test session shown above, the judge passed the task with this reasoning: "The agent successfully navigated to the specified URL, identified the page heading, scrolled to the actions table, and extracted the first five actions as requested." Turn the judge on when nobody watches the run: scheduled checks, unattended jobs and results that feed the next step of a workflow. Try it ``` Find the opening hours of the Louvre in a browser and have the judge check the answer. ``` ## Use cases with ready prompts Paste any of these into an assistant connected to FlyMy.AI and replace the example site with yours. ### 1. QA a website on mobile and desktop Open your pages at phone and desktop sizes, save screenshots and get a list of what looks broken; browser screens can be 320 to 6144 pixels wide. For a whole site, with every link checked and a published report page, use the [Website QA agent template](https://flymy.ai/templates/website-qa/). Try it ``` Open https://example.com in a browser at 390x844 and again at 1440x900. On each size, check the header, the main call to action and the footer, save a PNG screenshot of each view as a file, and give me a short report with the screenshots and what to fix first. ``` ### 2. Fill in forms in seconds with bu-ultrafast In our test, bu-ultrafast filled in the seven fields of httpbin.org's sample order form, submitted it and returned the server's echo in 11 seconds, for $0.0024. The agent is told to ask before purchases, sends or anything it cannot undo, so name the final step when you want it done. Try it ``` On bu-ultrafast, open https://httpbin.org/forms/post, fill in the pizza order form with test data (name Ada Test, size Medium, toppings Bacon and Extra Cheese, delivery 19:30) and submit it. Show me what the server echoed back. ``` ### 3. Research you can steer Comparisons take longer, and you often think of one more question halfway through. Start the task, add to it while it runs, and let the default model think harder on dense pages. Try it ``` In a browser with thinking_level high, find the monthly prices of the paid plans of 1Password, Bitwarden and Dashlane on their own pricing pages. While it runs, ask it to also check whether each one offers a family plan. ``` ### 4. Show a client what the agent did Send a read-only replay instead of screenshots pasted into a document: the client sees the task, every step and the answer, and you turn the link off after the review. Try it ``` Run a browser task that goes through the checkout of https://example.com up to the payment page without paying, then make a public read-only replay link I can send to my client. ``` ### 5. A weekly check with a verified result A scheduled agent can run the same browser task every week. With the judge on, a run passes only when the judge agrees, so a missing price or a broken button shows up as a failure instead of a cheerful summary. Try it ``` Create an agent that every Monday at 9:00 opens https://example.com/pricing in a browser, checks that every plan shows a price and a Sign up button, and runs the browser task with judge: true. Schedule it weekly and run it once now so I can see the result. ``` ## Browser Use MCP vs Playwright MCP vs computer use There are several ways to give an AI agent a browser. They differ in where the browser runs and in who plans each step. The table sticks to what each project documents; the sources are linked below it and were checked on 11 November 2026. | Approach | Browser Use MCP on FlyMy.AI | Playwright MCP | Claude computer use | Browser Use library | |---|---|---|---|---| | Where the browser runs | Browser Use Cloud | Where you run the server; the standard setup starts it with npx | A sandboxed computer you provide, such as a VM or container with a virtual display | A Browser Use Cloud browser by default, or a local one | | How the model works the page | Browser Use's agent runs the whole task from one instruction and returns the result | Your assistant calls one browser action at a time on accessibility snapshots, with no vision model needed | Claude reads screenshots and controls the mouse and keyboard | The model writes Python against the Chrome DevTools Protocol in a persistent REPL | | What you set up | One MCP URL and a sign-in | Node.js and an MCP config entry | The environment and the tool loop on the Claude API | Python 3.11+ and the browser-use package | | Keys | None, or your own Browser Use key | None for the server | A Claude API key | A Browser Use API key for the defaults, or your own LLM key | | License and price | Browser Use's own price, 0% markup, billed to your FlyMy.AI balance | Apache-2.0, free | Standard Claude tool use pricing | MIT; cloud browsers and models are paid | Sources: [Playwright MCP README](https://github.com/microsoft/playwright-mcp), [Claude computer use tool docs](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool), [Browser Use README](https://github.com/browser-use/browser-use), [FlyMy.AI Browser Use docs](https://docs.flymy.ai/mcp/browser-use). **How to choose.** Pick Browser Use MCP when you want a task done in the cloud from one instruction, with a live view, follow-ups and a shareable replay, and without running a browser yourself. Pick Playwright MCP when you want your assistant to drive a local browser step by step, for example while testing your own app. Pick computer use when the task goes beyond the browser into desktop applications. ## What it costs On FlyMy.AI, Browser Use usage on the shared account is billed to your FlyMy.AI balance at Browser Use's own reported price, with 0% markup: the browser time, the agent's model and any proxy traffic. With your own Browser Use key on the connection, the usage is paid on your Browser Use account instead. | Our test task, 11 November 2026 | Model | Time | Cost | |---|---|---|---| | Read the title of example.com | `bu-ultrafast` | 4 s | $0.0012 | | Fill in and submit a seven-field form | `bu-ultrafast` | 11 s | $0.0024 | | Read a page heading and a table, judge on | `gpt-6-luna` | 16 s | $0.011 | Times and costs are Browser Use Cloud's own figures for these sessions. Through an assistant, add a few seconds for the MCP round trip. Prices follow Browser Use's current rates. - `max_cost_usd` caps each task's model and search spend, $1 by default; browser time and proxy traffic come on top, and one session never goes past $10. - A browser kept open with `keep_alive` keeps costing until it closes, so stop the sessions you no longer need. ## FAQ ### What is Browser Use MCP? Browser Use MCP lets an AI assistant run tasks in a real cloud browser through the Model Context Protocol. Your assistant describes the task in plain words, Browser Use's agent does it in Browser Use Cloud, and the assistant gets back the output, the status, the cost and any files. ### Do I need a Browser Use API key? No. On FlyMy.AI, Browser Use runs on FlyMy.AI's shared account and is billed to your FlyMy.AI balance at Browser Use's own price, with 0% markup. With your own Browser Use plan, you can save your key on the connection instead. ### Is Browser Use free and open source? The Browser Use agent is open source under the MIT license and runs on any LLM, locally or on Browser Use Cloud, which is paid by usage. Through FlyMy.AI you pay Browser Use's own price per task, with no Browser Use subscription. ### How do I use Browser Use with Claude Code, Claude or Cursor? Add the FlyMy.AI MCP at https://mcp-agents.flymy.ai/mcp and sign in. In Claude Code, run claude mcp add --transport http flymyai https://mcp-agents.flymy.ai/mcp, then /mcp. In Claude, use Settings, Connectors, Add custom connector; in Cursor, add the URL as a remote MCP server. Then ask for a browser task. ### What is the difference between Browser Use MCP and Playwright MCP? Playwright MCP runs a browser where you run its server, and your assistant calls one browser action at a time on accessibility snapshots. Browser Use MCP hands the whole task to Browser Use's agent in the cloud and returns the result, with a live view, follow-ups, replay links and an optional judge. ### Which model should I use: bu-ultrafast, bu-fast or gpt-6-luna? Start with the default gpt-6-luna, which is economical and takes every thinking level up to max. Use bu-ultrafast for short interactive flows such as a form, and bu-fast when such a flow needs more reasoning per step. The Ultrafast presets stay fixed for the session. ### Can a browser agent log in to websites? Yes, with you in the loop. The person who starts a task gets a private live view and can take over for a password, a 2FA code or a CAPTCHA, then let the agent continue. A saved browser profile keeps the login for the next task. ### How much does a browser task cost? Browser Use's own price for the browser, the model and any proxy traffic, with 0% markup. In our tests, a page title on bu-ultrafast cost $0.0012, a seven-field form $0.0024 and a judged read on gpt-6-luna $0.011. Each task's model and search spend is capped by max_cost_usd, $1 by default, and one session never goes past $10. ### Is a public replay link safe to share? The replay is read-only: viewers see the task, the steps and the screenshots but cannot control the browser. They do see everything the session saw, including pages behind a login, so turn the link off when you are done. Never share the live view link, which controls the running browser. ## Give your agent a real browser Add FlyMy.AI to your assistant and run your first browser task in a minute. No Browser Use key, 0% markup on usage. [Connect Browser Use MCP](https://flymy.ai/mcp/browser-use/#connect) [Read the docs](https://docs.flymy.ai/mcp/browser-use) ## Keep reading - [Browser Use MCP Every action, setup commands and more requests.](https://flymy.ai/mcp/browser-use/) - [Website QA agent template A whole site on mobile and desktop, with a report page.](https://flymy.ai/templates/website-qa/) - [Browser agents that stay logged in Saved browser profiles for tasks behind a login.](https://flymy.ai/use-cases/browser-agents/) - [One MCP, 1,500+ tools Browser Use next to search, scraping, Google and Slack.](https://flymy.ai/mcp/) - [Platform log What is new on FlyMy.AI, release by release.](https://flymy.ai/changelog/) - [The agent loop is an interpreter. We compiled it. Our study of compiled agent programs.](https://flymy.ai/blog/compiling-ai-agents/) --- For agents: the FlyMyAI operating guide (sign up, connect, run models, tools and agents) is https://flymy.ai/skill.md, and https://flymy.ai/llms.txt lists the other pages and the documentation.