AI browser automation via a local server: your agent drives your real logged-in Chrome — you watch and can veto any tab.
Users
17
+14 in 30 days
Rating
0.0 / 5
unchanged in 30 days
Ratings
0
0 in 30 days
Store rank
#160K
#134K among extensions
Version
1.28.0
updated 16 Oct 2026
Users
monthly, 1 months
Latest, 16 Oct 2026:17 usersMove the mouse over the chart for any month
Ratings count
monthly, 1 months
Latest, 16 Oct 2026:0 ratingsMove the mouse over the chart for any month
The Chrome Web Store rounds user counts down to the leading digit, so the user line moves in steps. The ratings count is exact and is the better signal of real adoption.
About Chrome Bridge - AI Agent Browser Automation
chrome-bridge is browser automation for AI agents: ANY agent can drive the Chrome you're already using — your open tabs, your logged-in sessions, your SSO.
Playwright-style tools drive a browser they launched with a fresh profile; chrome-bridge drives your real browser. That's the only mode.
==
GETTING STARTED —3 steps, under a minute:
1. Install this extension.
2. Click its toolbar button and press "Copy setup prompt for your AI agent".
3. Paste that text to your agent (Claude Code, Cursor, or any agent that can run shell commands).
The agent does the rest itself: downloads the tiny zero-dependency bridge, starts it, verifies the connection, and runs a no-trace dry run before reporting back. The popup's dot shows when the bridge is ready.
==
WHAT YOU GET
• Your agent can read pages, click, fill forms, navigate, screenshot, capture network traffic — in YOUR logged-in browser.
• AI-first under the hood: the agent reads pages as a compact accessibility tree it can narrow to one region or diff against its last look, instead of re-reading screenshots, and every command is a one-line shell call any agent can run.
• Full local control: everything runs on 127.0.0.1, nothing leaves your machine. No cloud relay, no account, no telemetry.
• Built for the human watching: every tab the agent touches wears a 🟣 pill narrating each action in real time, joins a 🟣 tab group, and shows status in the favicon; every action lands in a history you can review; one trusted click on ⏏ rejects the agent from that tab.
• Point the agent at an exact tab: the toolbar popup copies a tab reference (id:…) that pins every command to that tab.
• Honest actions: if the page re-sorts under a saved element reference, the action fails loudly instead of clicking the wrong element; a click the app ignored says so instead of faking success; trusted input refuses to fake-fire on a hidden tab; a fill an editor's autosave can't see carries a warning; typing into heavy editors (docs, block editors) is paced and verified.
• Rich editor savvy: the agent can drop whole structured documents (headings, lists, links) into block editors in one shot — and clear them again, even when the editor owns its content model.
• No residue: a `doctor` command finds and cleans any leftover markers, and dead-session markers are reaped automatically on browser restart.
• Multiple Chrome profiles supported — the agent never silently acts in your personal browser when you meant the work one.
Advanced users can also run it from source (git clone) — see the GitHub repo.
==
PRIVACY
The extension talks only to a local server on 127.0.0.1 that you (or your agent) start. No data leaves your machine.
Full policy: https://raw.githubusercontent.com/siropkin/chrome-bridge/master/PRIVACY.md