QAmcp

Copied. Paste it into Claude Code, Cursor or Codex:Fetch and follow the instructions at https://qamcp.dev/setup.md to test my MCP server.

Your AI finds your server in your project, tests it with QAmcp, explains what's wrong in plain words, fixes it with your OK, and tests again. Read the instructions

SoonYour score can change any day. QAmcp Monitor tests your server every hour and emails you when it breaks.See what Monitor does →

Directory / Browser Control MCP by Nova (CIVAI)

Browser Control MCP by Nova (CIVAI)

Q: Is Browser Control MCP by Nova (CIVAI) server safe?

QAmcp's test on 2 Nov 2026: None of the hidden-instruction wordings we know were found in its tool descriptions. 0 of its 24 tools are labeled read-only; the others may change or send things. Overall it scored 74/100 on 2 Nov 2026. A test can only see what the server shows from outside, not what it does with your data behind the scenes. Connect servers from makers you trust.

by nova.civai.co · https://nova.civai.co/mcp/agents/mcp-browsergpt-browser-control

Control your Chrome from chat: open websites, read what's on the page, click buttons, and fill in f…

24 tools✓ no account neededPricing: not listed

Visit website

74 /100

PASS · WITH NOTES

Partial: 9 of 12 checks · 2 Nov 2026

Is Browser Control MCP by Nova (CIVAI) server safe?

None of the hidden-instruction wordings we know were found in its tool descriptions.
0 of its 24 tools are labeled read-only; the others may change or send things.
Overall it scored 74/100 on 2 Nov 2026.

A test can only see what the server shows from outside, not what it does with your data behind the scenes. Connect servers from makers you trust.

What it can do

24 tools, in the server's own words.

set_browsergpt_license

Save BrowserGPT license when MCP headers did not already provide it (Authorization Bearer license).

list_browsers

List BrowserGPT browser nodes (local + remote) for the configured license.

ensure_remote_runner

Start/prepare Nova remote BrowserGPT runner (ensure-remote-browser).

open_control_session

Open MCP control session on target browser node_id.

browser_tabs

List tabs in the open control session.

browser_snapshot

Accessibility snapshot with refs.

browser_screenshot

JPEG screenshot of the session viewport.

browser_navigate

Navigate a tab to URL.

browser_get_cookies

Read session data for a tab via BrowserGPT extension: chrome.cookies (HttpOnly), optional localStorage/sessionStorage keys.

browser_exec_js

Run JavaScript on a target tab (same CDP evaluate path as BrowserGPT Quick Task exec_js).

browser_actions

Low-level action batch (JSON array string).

browser_interact

Interact via snapshot ref.

list_input_methods

List BrowserGPT dom input method ids; optional page_url for domain-specific hint.

browser_click

Click by snapshot ref (Cursor parity).

browser_fill

Fill field by ref.

browser_type

Type into field by ref.

browser_press_key

Press keyboard key in tab.

browser_scroll

Scroll wheel; optional ref scrolls element into view first.

browser_select_option

Select options on a select ref.

browser_drag

Drag between viewport coordinates.

browser_mouse_click_xy

Click at viewport x/y when snapshot refs are empty.

browser_get_bounding_box

Get ref bounding box from last snapshot.

browser_highlight

Briefly outline ref on page.

browser_dom_input

Run dom input method chain (e.g.

Test results

Safe mode: only tools labeled read-only were run. How the test works

!
Ready for the newest MCP version
1 warning
- !Uses an older MCP version (2025-11-25). It works today, but it isn't on the July 2026 version (2026-07-28) yet.
note
!
Every tool is described
17 warnings
- !browser_snapshot: Inputs with no description: interactive_only
- !browser_navigate: Inputs with no description: url
- !browser_actions: Inputs with no description: actions_json, timeout_ms
- !browser_interact: Inputs with no description: tab_id, op, ref, text, key, code, input_method
- !list_input_methods: Inputs with no description: page_url
- !browser_click: Inputs with no description: tab_id, ref, element
- !browser_fill: Description is very short (18 characters)
- !browser_fill: Inputs with no description: tab_id, ref, value, input_method
- and 9 more
note
!
Tools are labeled safe or dangerous
1 warning
- !24 of 24 tools have no safety label (read-only or dangerous): set_browsergpt_license, list_browsers, ensure_remote_runner, open_control_session, browser_tabs, browser_snapshot, browser_screenshot, browser_navigate, browser_get_cookies, brow
note
!
Follows MCP best practices (not scored yet)
2 tips (not scored)
- !24 tools not labeled read-only don't say whether running them twice is safe (idempotentHint): set_browsergpt_license, list_browsers, ensure_remote_runner, open_control_session, browser_tabs, browser_snapshot and 18 more
- !24 tools don't say whether they reach outside services like the web (openWorldHint): set_browsergpt_license, list_browsers, ensure_remote_runner, open_control_session, browser_tabs, browser_snapshot and 18 more
note
!
Resources and prompts work
1 warning (3 resources)
- !civai-secops://browsergpt-browser-control/@public/CAPTCHA_AGENT.md: The content contains "don't tell the user". If an AI reads this, it may follow it. Check where this text comes from.
- and 2 more
note
✓
ConnectsConnected in 202 ms
pass
✓
Lists its toolsFound 24 tools
pass
✓
No known hidden instructionsNone of the hidden-instruction wordings we know were found in tool descriptions. New wordings can get past this check.
pass
✓
Fast enough
All good
- ✓Connecting took 202 ms
- ✓Listing tools took 140 ms
pass
–
Tools work with good inputs
Not tested
- –set_browsergpt_license: Not run: no safety label (if you own this server, test again in owner mode)
- –list_browsers: Not run: no safety label (if you own this server, test again in owner mode)
- –ensure_remote_runner: Not run: no safety label (if you own this server, test again in owner mode)
- –open_control_session: Not run: no safety label (if you own this server, test again in owner mode)
- –browser_tabs: Not run: no safety label (if you own this server, test again in owner mode)
- –browser_snapshot: Not run: no safety label (if you own this server, test again in owner mode)
- –browser_screenshot: Not run: no safety label (if you own this server, test again in owner mode)
- –browser_navigate: Not run: no safety label (if you own this server, test again in owner mode)
- and 16 more
not tested
–
Bad inputs give a clean errorNot tested
not tested
–
Tools work togetherNo safe tool returned data to start a chain from, so chains were not tested (not counted either way)
not tested
–
No secrets leakedNo error messages to check
not tested

Test it yourself

Free and open source. Same checks, same score.

npx qamcp-cli https://nova.civai.co/mcp/agents/mcp-browsergpt-browser-control

Show this score on GitHub

Free for every server in the directory. Paste this into your README: the badge shows the score from our latest test.

[![QAmcp score](https://qamcp.dev/s/co-civai-nova-mcp-browsergpt-browser-control.svg)](https://qamcp.dev/s/co-civai-nova-mcp-browsergpt-browser-control.html)

Score over time

2 Nov 202674

One test so far. New tests will show here.

Reviews

Coming soon: reviews from people who use Browser Control MCP by Nova (CIVAI), signed in with GitHub, one per person. Reviews never change the score.

Is this your server? Get an alert the moment it breaks.

QAmcp Monitor (coming soon) tests it every hour and emails you when it goes down, the score drops, a tool changes or a secret leaks. You'll also be able to claim this page and get a verified owner mark. Founding price: $19/month for life for the first 50 teams.

Have a concern, or want it removed? Ask us and we'll take it off the list. How the directory works