# Scrappy > Scrappy (scrappy.gg) is a cloud browser: real, headful Chrome on residential connections that code, an AI agent or a person can drive. Saved logins replay from their own fixed IP, a person can take over a live session from a phone, and the platform is reachable over a REST API, a hosted MCP server, or Claude's browser tool. A lead pipeline (scraping, email verification, outreach) runs on the same account. Pay as you go: a session is 5 credits to start, then 1 per minute; new accounts get 100 credits. Every link below is the page as markdown. All of it in one file: https://scrappy.gg/llms-full.txt ## Product - [Cloud browser](https://scrappy.gg/md/cloud-browser): what a session can do - [Pricing](https://scrappy.gg/md/pricing): credit packs and what a browser hour costs - [Compare](https://scrappy.gg/md/compare): feature matrix against hosted browsers and lead tools ## Questions - [Keep a Puppeteer session logged in](https://scrappy.gg/md/questions/keep-puppeteer-session-logged-in): To keep Puppeteer or Playwright signed in across runs, save cookies and local storage, restore them before the first page load, and reuse the same IP. - [How to give Claude a web browser](https://scrappy.gg/md/questions/give-claude-a-browser): Give Claude a browser through its own browser toolset or an MCP server. Either way, something must run a real browser and return what it sees. - [Hosted browser with residential proxy](https://scrappy.gg/md/questions/hosted-browser-with-residential-proxy): Most hosted browsers bill residential proxy traffic per GB on top of browser time. Scrappy includes the residential exit in the per-minute price. - [Why browser automation gets blocked](https://scrappy.gg/md/questions/why-browser-automation-gets-blocked): Automation is blocked by headless mode, datacenter or flagged IPs, and a browser whose timezone disagrees with its IP — before any fingerprint patch matters. - [Scrape a website that requires a login](https://scrappy.gg/md/questions/scrape-website-that-requires-login): To scrape behind a login without being blocked, sign in once as a person, save the session, and replay it in real Chrome from the same residential IP. - [Automate a login that uses two-factor](https://scrappy.gg/md/questions/automate-login-with-two-factor): Don't script the second factor. Let a person finish two-factor sign-in once in a live browser, then save the session so later runs start already signed in. - [Know when an automated login expires](https://scrappy.gg/md/questions/know-when-automated-login-expires): An expired login shows up as a sign-in page, a password field or a vanished auth cookie. Check for those, or have a webhook tell you as it happens. - [Let a person take over a cloud browser](https://scrappy.gg/md/questions/let-a-person-take-over-a-cloud-browser): Open the session's live viewer: a person can watch, click and type in the same browser the automation drives, then hand it back. Scrappy's works on a phone. - [MCP server for browser automation](https://scrappy.gg/md/questions/mcp-server-for-browser-automation): A browser MCP server gives an AI agent tools to open pages, click and read them. Scrappy's is hosted: one URL, one API key, 62 browser tools on real Chrome. ## Comparisons, use cases and integrations - [Browserbase alternative](https://scrappy.gg/md/alternatives/browserbase): Scrappy as a Browserbase alternative: real headful Chrome with the residential proxy in the price, saved logins on a fixed IP, and takeover from a phone. - [Steel alternative](https://scrappy.gg/md/alternatives/steel): Scrappy as a Steel alternative: real headful Chrome with the residential proxy in the price, saved logins on a fixed IP, and takeover from a phone. - [Browserless alternative](https://scrappy.gg/md/alternatives/browserless): Scrappy as a Browserless alternative: real headful Chrome with the residential proxy in the price, saved logins on a fixed IP, and takeover from a phone. - [Hyperbrowser alternative](https://scrappy.gg/md/alternatives/hyperbrowser): Scrappy as a Hyperbrowser alternative: real headful Chrome with the residential proxy in the price, saved logins on a fixed IP, and takeover from a phone. - [Anchor alternative](https://scrappy.gg/md/alternatives/anchor): Scrappy as a Anchor alternative: real headful Chrome with the residential proxy in the price, saved logins on a fixed IP, and takeover from a phone. - [Scrape websites behind a login](https://scrappy.gg/md/use-cases/scrape-behind-login): Automate sites that need a sign-in: capture the login once, replay it from the same residential IP, and get a webhook when the site ends it. - [A browser for AI agents](https://scrappy.gg/md/use-cases/browser-for-ai-agents): Give your AI agent a real browser: 57 MCP tools, Claude's browser tool, approval before risky steps, and a person who can take over live. - [Human-in-the-loop browser automation](https://scrappy.gg/md/use-cases/human-in-the-loop-browser): Automation that hands off to a person: take over a cloud browser from a phone, clear the sign-in wall or CAPTCHA, and hand it back. - [Claude browser tool, hosted](https://scrappy.gg/md/integrations/claude): Run Claude's browser toolset against a real cloud browser: post the assistant's tool calls to one endpoint and send back the results. - [Browser MCP server](https://scrappy.gg/md/integrations/mcp): A hosted MCP server with 57 cloud-browser tools. Connect Claude Code or any MCP client with one endpoint and one API key. ## Articles - [Browserbase Alternatives for AI Agents: What Actually Differs (2026)](https://scrappy.gg/md/blog/browserbase-alternatives-for-ai-agents): Anchor, Steel, Hyperbrowser, Browserless and Scrappy compared on what changes the bill and the block rate: proxy pricing, fixed IPs per login, and takeover. - [Choosing an MCP Server for Browser Automation: What Matters Behind the Tools](https://scrappy.gg/md/blog/choosing-a-browser-mcp-server): Browser MCP servers expose similar tools. What decides whether your agent gets through is the browser behind them: headful, residential, logged in, watchable. - [How to Give Claude (or Any AI Agent) a Real Web Browser](https://scrappy.gg/md/blog/give-claude-a-real-browser): Three ways to let Claude drive a browser — an MCP server, Claude's own browser toolset, or a hosted agent run — with working code and what breaks in practice. - [A Hosted Browser With the Residential Proxy Included: How It Works and What It Costs](https://scrappy.gg/md/blog/hosted-browser-residential-proxy-included): Why signed-in automation needs a residential IP, why most hosted browsers bill it per GB, and how we include a fixed residential exit in the session price. - [How to Keep a Playwright or Puppeteer Session Logged In Across Runs](https://scrappy.gg/md/blog/keep-browser-session-logged-in): Save storage state, restore it before the first page load, and return from the same IP. The local recipe in code, four ways it still fails, and the hosted way. - [When Your Browser Agent Gets Stuck, Hand the Browser to a Person](https://scrappy.gg/md/blog/hand-stuck-browser-agent-to-a-human): CAPTCHAs, two-factor codes and press-and-hold walls stop every browser agent. Detect the wall, text a person a live link, and carry on in the same session. - [How to Scrape a Website That Requires a Login (Step by Step, With Code)](https://scrappy.gg/md/blog/scrape-website-that-requires-login): Sign in once in a real browser, save the login, and replay it from the same IP on every run — with working Python, and what to do when the site signs you out. - [A JavaScript Dialog Looks Exactly Like a Frozen Tab](https://scrappy.gg/md/blog/dialog-looks-like-frozen-tab): Every CDP command on the tab times out, CPU is at zero and the screen shows a normal page. Before you call it a hung renderer, rule out a native dialog. - [Your Proxy Provider Replaced the IP. Your Config Still Has the Old One.](https://scrappy.gg/md/blog/proxy-ip-replaced-silently): A static residential IP was replaced overnight, the old address answered 407, and every login pinned to it failed. What we changed so it cannot recur. - [puppeteer.connect() Hangs Forever. Here Is Why, and How to Tell.](https://scrappy.gg/md/blog/puppeteer-connect-hangs): A puppeteer connect that never resolves while raw CDP answers instantly is usually one target that never attached, not a dead browser. How to tell. - [Headless Chrome Is Why You're Getting Blocked](https://scrappy.gg/md/blog/headless-chrome-is-why-youre-blocked): You can patch every fingerprint you know about and still be detected, because headless is visible on its own. What we changed after chasing this for months. - [Your Scraper Can't Log In. That's Where the Good Data Is.](https://scrappy.gg/md/blog/scraper-cant-log-in): Everything easy to scrape has already been scraped. The contacts worth having sit behind a sign-in — and that is a different engineering problem. ## Docs - [API reference](https://scrappy.gg/md/docs/api) - [Quick start](https://scrappy.gg/md/docs/guides/quickstart): From sign-up to your first leads in under five minutes - [Dashboard overview](https://scrappy.gg/md/docs/guides/dashboard): Find your way around the dashboard - [Cloud browser](https://scrappy.gg/md/docs/guides/cloud-browser): Sessions, the live viewer, actions, batches, files and recordings - [Saved logins](https://scrappy.gg/md/docs/guides/browser-profiles): Profiles that keep a sign-in, pinned to their own exit IP - [Claude's browser tool](https://scrappy.gg/md/docs/guides/claude-browser-tool): Run Claude's browser toolset against a session - [MCP server](https://scrappy.gg/md/docs/guides/mcp): Give your own AI agent the platform as tools - [Webhooks and integrations](https://scrappy.gg/md/docs/guides/webhooks): Signed events, delivery history, Gmail and data destinations - [Credentials](https://scrappy.gg/md/docs/guides/credentials): Store logins for sites that need one, encrypted - [Teams and API keys](https://scrappy.gg/md/docs/guides/enterprise): Organizations, scoped keys, templates and proxy profiles - [Credits and billing](https://scrappy.gg/md/docs/guides/billing): What each action costs, and how to buy more - [Privacy and GDPR](https://scrappy.gg/md/docs/guides/privacy): Consent, data export and account deletion