Ask any page
"Is this contract safe to sign?" "What does the third paragraph mean?" Claude reads the full page and answers in context.
Try a 4,000-word article →Ask, summarize, fill forms. All without leaving the tab. Claude already knows you. Now it lives on every page.
⌘ + ⇧ + Space to open. Anywhere. Any page.
"Is this contract safe to sign?" "What does the third paragraph mean?" Claude reads the full page and answers in context.
Try a 4,000-word article →Job applications, surveys, contact forms, registrations. Claude pulls from what it already knows about you and fills every field in one keystroke.
Try a real job application →Wearable sensors detect subclinical atrial fibrillation ✦ ← hover to expand episodes that conventional ECGs miss.
Select any text and a small ✦ dot appears, faded into the background. Hover to expand: Ask, Summarize, or Explain. Never a popup blocking your reading.
Try it on a dense article →"Compare pricing across these three pages." Reference multiple open tabs in one prompt. Claude reads them all and answers as one.
Try a 3-page comparison →Not a token reseller. Not a middleman. Not an analytics box.
You select text, open the palette, fill a form. The extension reads only what you act on.
The extension uses your existing claude.ai cookie. No API key. No re-auth. No new account.
Streaming back to your tab. With full page context. Without ever leaving where you are.
No analytics. No third-party servers. No data resale. Read the privacy policy if you want every detail.
Yes. Browse with Claude uses your existing claude.ai session: Pro, Team, or Enterprise. We don't take an API key, we don't bill you, we don't resell your tokens. The only cost is whatever Claude already costs you.
Page content (only when you open the palette or reference a tab), text you select, form-field labels when you ask Claude to fill them. Everything goes from your browser straight to your authenticated claude.ai session, with the same privacy as if you pasted it there yourself.
Yes. We detect canvas-based apps and read your selection via the clipboard. For pages where DOM extraction yields little text, a viewport screenshot is sent so Claude can read visually.
claude.ai is the conversation. Browse with Claude is the place: it lives on the page you're reading, the form you're filling, the article you're skimming, without you having to switch tabs.
One click. Nine fields. Claude pulls it all from what it already knows about you.