perfect working very well with codex and claude code!, i am shocked there are only 1k users lmao!
Agent360 Browser MCP
- 2,000+ users
- 5.004 ratings
- #1,548in Developer Tools
Your agent works in the Chrome you're signed into and can ask you for a 2FA code or a decision. 40 tools, MCP clients, MIT.
by Agent360 Group ApS · 1 listing
- Users
- 2,000+
- #1,548 of 31,410 in category
- Rating
- 5.00
- 4 ratings · median 4.63
- Version
- 1.31.0
- 178 KiB · Manifest V3
- Access
- High75/100
- 10 permissions + every website
Description
As written by the publisher on the storeBrowser MCP gives your AI agent - any MCP client that can run a local server - control
of the Chrome you already use - the one with your sessions, your cookies, your logins.
Not a fresh headless browser that hits a login wall on the first page.
⚠ READ FIRST - THIS EXTENSION IS HALF OF BROWSER MCP
Browser MCP is a Chrome extension PLUS a local MCP server. This page can only give
you the extension. On its own it will sit on "Not connected" forever, because there
is no server for it to talk to. The other half takes one paste.
═══ SETUP ═══
1. Install this extension (you are here).
2. Ask your AI assistant to set up the other half. Paste this in your AI client:
----------------------------------------------------------------
Set up Browser MCP for me. Read https://browsermcp.dev/llms-install.md
and follow it for whichever AI client you are running in. One step
only I can do - tell me which one, and wait for me. When I confirm,
call browser_list_tabs to prove both halves are talking.
If you cannot open URLs: npx @agent360/browser-mcp install --skip-extension
(it registers the server in the MCP clients it finds on this machine; in
other clients, add it by hand as below)
----------------------------------------------------------------
- Restart your AI client when it tells you to. A green badge appears on the extension's icon
- the first time the browser is actually used - no badge before that is normal.
Rather do it by hand? Add the server to your MCP client's configuration:
{"mcpServers": {"browser-mcp": {"command": "npx",
"args": ["@agent360/browser-mcp@latest"]}}}
(Some clients name the root key "servers"; command-line clients have a one-line
"mcp add" command.) Step-by-step guide for each client: https://browsermcp.dev/#install
Why two steps? Chrome does not allow an extension to install itself from npm, and
npm cannot install a Chrome extension. Neither half can install the other, so you
install each once.
Stuck on "Not connected"? https://browsermcp.dev/docs/troubleshooting/
═══ WHAT TO SAY ONCE IT WORKS ═══
Your agent does nothing on your pages until you ask it to. Paste one of these
(put your own site where it says example.com):
"Open example.com and take a screenshot."
Start here - a screenshot back means both halves are talking.
"Open Gmail and tell me who sent my last 3 emails."
The one that shows the difference: it works because it is YOUR
browser, already signed in. A fresh browser without your login would
hit a login wall here.
"Go to my analytics dashboard, pull this month's numbers, and put
them in a table."
A dashboard you are already logged into, when the numbers are on the
page - no API key, no export.
"Fill in the signup form at example.com/signup with my details. Stop
and ask me before anything sensitive."
Your agent can stop and ask you in the page before passwords or
payment details.
"Log me in at app.example.com. If it emails a code, open Gmail, read
the code and continue."
No email API to set up - it reads the code the way you would.
The pattern: many of the browser tasks you would do yourself, on sites you
are already signed into. Strongest where there is no API - internal
dashboards, admin panels, portals.
═══ WHAT IT CAN DO ═══
40 tools, including
• Navigate, click, double-click, right-click, hover, scroll, press keys
• Read page content, extract lists, take screenshots, read console logs
• Fill forms, including React-controlled inputs, dropdowns, comboboxes,
date pickers and file uploads
• Work inside iframes, switch and close tabs, wait for network activity
• Run your agent's own JavaScript in the page or in one of its frames -
only through Chrome's debugger, which Chrome announces with its own
notice while it is attached
• Read cookies for the sites it has open, set cookies, read and write
localStorage
• Open the API page of nine common services, with notes on where to
find or create the key
• Answer alerts, confirms and prompts. If a page asks "Leave site?"
while the agent navigates away, the answer is Stay by default and your
agent is told so; the tool tells it to ask you before it calls again
with leave_page:true, which leaves and loses unsaved work. An armed
browser_handle_dialog answers instead of either
• Detect CAPTCHAs (reCAPTCHA v2/v3, hCaptcha, Cloudflare Turnstile,
FunCaptcha): it tries the reCAPTCHA checkbox; for an image grid your
agent takes a screenshot and picks the cells; if that does not work,
your agent can ask you to finish it in the tab
• Your agent can ask you directly - an in-page dialog for 2FA codes,
passwords and anything else it should not decide alone. The page
cannot read what you type in it, but it can see the keystrokes, so
type only secrets that belong to the page the tab is on
═══ TWENTY AGENTS, ONE BROWSER ═══
Up to 20 agent sessions can use one Chrome at the same time. Each MCP server
gets its own named, colour-coded tab group (Claude 1, Claude 2 ...) - except a
tab it opens in a window of its own, which is not grouped - can only see its
own tabs, and keeps up to 20 of them open: opening more closes its oldest
tabs, but never the one it is working in or the one it just opened. Clients
that start one server per chat get one group per chat;
chats that share a server share its group. When the server goes away, its
tabs close with it.
═══ MORE THAN ONE CHROME PROFILE ═══
With the extension in several Chrome profiles, give each profile a name in
the extension's popup ("This Chrome profile"), and set BROWSER_MCP_PROFILE=<name>
in a client's MCP config to make that client drive that profile.
A chat keeps using the chosen installation while it remains available. If it
disconnects, a call reports the loss and sends nothing. A later call chooses
again, using BROWSER_MCP_PROFILE if configured.
The name is a label to choose by, not a password.
═══ WHAT IT DOES WITH YOUR DATA ═══
Nothing is sent to us. The extension takes its orders only from a Browser MCP
server on 127.0.0.1 - your own computer - normally the one your AI client
started. The pages your agent opens, and the HTTPS calls
of browser_fetch, go to the sites you point it at. There is no account, no telemetry, no server of ours,
and no payment. What your agent reads goes to your AI client and its model
provider, the same as anything else you show it. The profile name and a random
installation id are part of the answers your agent gets, so they reach your AI
client too. The extension keeps a short local log of the agent's
last 50 actions (its popup shows the latest 30), with cookie and storage reads
(get_cookies, get_local_storage) and the agent's own JavaScript (execute_script,
select_frame) highlighted, so you can look at what happened; you can clear it
from the popup.
═══ WHAT IT CANNOT DO ═══
In a tab you are not looking at, the agent can navigate, read, screenshot,
run scripts, click, and fill a field it finds by CSS selector. It cannot press
keys, hover, double-click, right-click, click by coordinate, type into a
combobox, or fill a field found by its visible text - Chrome does not deliver
mouse and keyboard events to a tab that is not the visible one in its window.
When one of those is needed, the tool says so instead of failing silently,
and, after asking you, the agent can bring the tab and Chrome in front of you with browser_switch_tab
(bring_to_front). Without that, browser_switch_tab does not bring Chrome's
window forward. (browser_screenshot has the same bring_to_front option, and
browser_ask_user makes its tab the active one in its window and restores a
minimized window, but does not bring Chrome in front of other apps.) This is a
Chrome limitation, not a bug we are about to fix.
Two more limits of a tab you are not looking at: a screenshot needs Chrome's
window to be visible - if it is covered by other windows, the answer is
"window-covered" and nothing is taken - and a list that loads its rows as you
scroll can come back incomplete in a hidden tab; the answer then says it did
not reach the end.
MIT licensed, source at https://github.com/Agent360dk/browser-mcp
Ratings and reviews
Split of the 4 written reviews we hold4 written reviews we hold
This plugin is freaking awesome. It absolutely blows that abandoned Browser MCP out of the water. Why does it have so few users?
Track Agent360 Browser MCP on Radar
This page shows the listing as it is today. Add it to Radar and get one email when it changes.
- Users and rating
- Store text and screenshots
- Permissions it asks for

Agent360 Browser MCP, in plain words
What it is, what to use instead, what it can access and how to fix it.
Agent360 Browser MCP at a glance
Agent360 Browser MCP is a Developer Tools extension by Agent360 Group ApS with 2,000 users. Chrome Web Store users rate it 5.00 out of 5 from 4 ratings, which is still a small sample. By users it sits in the top 5% of Developer Tools extensions.
- By users
- #1,548of 31,410 in Developer Tools
- Rating
- 5.00category median 4.63
- Access
- High75 of 100
Agent360 Browser MCP alternatives
Closest by what they do · users as the store shows them · bars on a log scaleIs Agent360 Browser MCP safe?
Agent360 Browser MCP asks for 10 permissions and access to every website. 5 of them give wide access.
Permissions show what it could do, not what it does. See the full permission list.
<all_urls>Read and change data on every websitetabsSee the address and title of every open tabcookiesRead and change site cookies, including loginswebNavigationFollow every page you opendebuggerTake full control of pages through the debugger
Every permission Agent360 Browser MCP asks for
10 permissions + every website · 5 sensitive · High access overallSites
- Read and change data on every website
<all_urls>
Browser permissions · 10
- See the address and title of every open tab
tabs - Read and change site cookies, including logins
cookies - Follow every page you open
webNavigation - Take full control of pages through the debugger
debugger - Organize tab groups
tabGroups - Run its code on pages it can access
scripting - Save its own settings
storage - Run tasks on a schedule
alarms - Run hidden helper pages
offscreen - Show desktop notifications
notifications
Sensitive marks access that reaches your data or other sites. The access level is a 0 to 100 score of all of it: under 25 is Low, 60 and up is High.
Agent360 Browser MCP not working?
What recent reviews say
Store reviews from the last 90 daysCommon fixes
Try them in this order- 1Update Chrome and the extension
- 2Turn it off and on in chrome://extensions
- 3Reload the page
- 4Turn other extensions off in case two clash
Still stuck? Contact the maker or try an alternative.
Know when a listing changes
Track any extension on Radar: users, rating, permissions and store text, with one email when something moves.