Browsr Relay

Browsr Relay
One-click browser control for your AI agents.
Turn any Chrome tab into a live, controllable session for your AI
coding agents. Browsr Relay bridges your browser directly to
Claude, Codex, and OpenClaw — no VMs, no headless browsers, no
setup.

How it works

1. Sign in with your Browsr account
2. Click "Start Relay" on any tab
3. Your agents can now see and control your browser — navigating,
clicking, typing, and extracting data in real time
That's it. Your existing browser session, cookies, logins, and
extensions all stay intact.

Why Relay?

Traditional browser automation spins up isolated, sterile browser
instances that can't access your logged-in sessions or
authenticated apps. Relay flips that — your agents operate your
actual browser, seeing exactly what you see.

- Works with Claude — give Claude eyes and hands on the web
- Works with Codex — let Codex research, test, and verify in a
real browser
- Works with OpenClaw — connect your autonomous agents to live
web sessions
What your agents can do
Navigate
Navigate to URLs, go back, go forward, refresh — full browser
history control.

Interact
Click elements, type text, press keys, hover, focus, select
dropdowns, check boxes, clear fields — everything a human can do.
Extract
Get page text, titles, HTML content, element attributes, or run
arbitrary JavaScript with evaluate.

Observe
Full DOM snapshots with an interactive element map — your agent
sees every clickable, typeable, and selectable element on the
page.
Screenshot
Viewport or full-page captures, plus a live screenshot stream so
your agent watches the page update in real time.

Wait
Wait for elements to appear or navigation to complete — proper
synchronization so your agent doesn't race ahead of the page.

Mouse & Keyboard
Low-level input when you need precision: move mouse, scroll to
coordinates, drag and drop between elements.
Scrape & Crawl
Extract markdown, HTML, links, images, or structured JSON from
any URL. Crawl multi-page sites with depth and path controls.
Search
Web search with optional full-page scraping of results.
Shell
Sandboxed code execution containers — run code alongside your
browser automation.

Full CDP Access
Every Chrome DevTools Protocol method is available. No
allowlists, no restrictions. If Chrome can do it, your agent can
do it through Relay.
Ready-to-use skills for your agents

Download pre-built skills from https://browsr.dev/docs/skills/
and drop them into Claude, Codex, or OpenClaw. Your agent gets
browser superpowers in seconds — no configuration, no prompt
engineering.

Features

- One-click connect — start and stop relay with a single button
- Live screenshots — agents see your browser in real time with
configurable capture rates
- Full browser control — the complete CDP protocol, unrestricted
- Secure authentication — passwordless OTP login, JWT-based
sessions
- Real browser context — your cookies, extensions, and logged-in
sessions are all available
- Lightweight — no background resource usage when relay is off
Built for agents
Let your agent instantly see your screen. Start a relay and your
agent is looking at exactly what you're looking at — whether
you're a developer debugging a production issue, a product
manager researching competitors, or a designer reviewing a live
site. One click and your agent has full context, no copy-pasting
URLs or describing what's on screen.

Privacy & transparency

You're always in control and always aware. Relay shows a clear
active indicator on the tab being captured — you'll never wonder
which tab your agent can see. Only the specific tab you choose is
relayed, and only while you have it turned on. Screenshot
capture and control stop immediately when you click "Stop Relay"
or close the extension. No background activity, no silent
capture, no surprises.

Documentation

Full API reference, setup guides, and integration examples at
https://browsr.dev/docs/.