Browser Tools MCP logo

Browser Tools MCP

Inspect a real Chrome session through MCP with console, network, tab, screenshot, storage, and Lighthouse tools.

Install Browser Tools MCP

Pick your client, then add it in one click or paste the config.

STDIONode.js
~/Library/Application Support/Claude/claude_desktop_config.json
claude_desktop_config.json
{
  "mcpServers": {
    "browser-tools-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@agentdeskai/browser-tools-mcp"
      ]
    }
  }
}

Settings → Developer → Edit Config, merge this into mcpServers, then fully quit and reopen Claude Desktop.

Automated check passed— started and listed 16 tools correctly (10d ago).

Overview

browser-tools-mcp connects MCP-compatible clients to an existing Chrome session through a DevTools extension. It exposes console and network telemetry, page and tab inspection, screenshots, browser actions, storage reads, and Lighthouse audits without requiring a separate browser profile. Reach for it when an agent needs to debug the authenticated page already open in Chrome rather than drive a fresh automated browser. It runs locally over loopback and supports structured MCP outputs and resource links for larger results.

Use cases

  • Inspect console errors from an authenticated Chrome session
  • Trace failed API requests and export captured network traffic
  • Run Lighthouse audits against the page currently under review
  • Capture a screenshot and inspect the selected DOM element
  • Reproduce a clean issue after clearing browser telemetry

Key features

  • Console and network telemetry with filters and paging
  • Inspection of existing Chrome tabs and page content
  • Screenshots, browser storage reads, and tab navigation
  • Lighthouse accessibility, performance, SEO, and best-practices audits
  • MCP resources for full logs, HAR files, screenshots, and audit reports
  • Loopback-only connector with structured MCP output schemas

Tools16

What Browser Tools MCP lets your agent do. Expand a tool for its parameters.

Verified live~3.1k tokens of context

Security & health

What it can reach and how reliably it runs — from our own checks, not the maintainer’s claims.

ActiveLast checked Today

Recent health check succeeded.

18/18 checks healthy over the last 11d

What it can do

From the 16 tools this server returned when we ran it; 16 declare their own safety hints.

Can delete or modify data · 1Handles credentials · 2Makes network requests · 6Read-only tools · 15
Network at startup3 connections
Files written at startup1
Install scriptsNone
npm provenanceNot published
Package age620 days
Maintainers1

Observed 2026-10-27 in an isolated sandbox. A description of reach, not a security audit.

Supply-chain signal

No high-severity advisories surfaced by our automated scan.

Critical 0High 0Medium 0Low 0

Scanned 11d ago via OSV.dev · @agentdeskai/browser-tools-mcp (npm)

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks — not a rating.

GitHub stars7.3kStargazers on the source repository.
npm downloads17kPackage downloads in the last 30 days.
Last commit9d agoMost recent push to the default branch.
Availability100%Our rolling endpoint + install checks that succeeded.
Install checkPassedOur sandbox started it and listed its tools.
Tools exposed16Callable tools this server registers over MCP.
Directory activity3 viewsConfig copies, upvotes, and views on AllMCPs.

How Browser Tools MCP works

What browser-tools-mcp does

browser-tools-mcp gives an MCP client access to an existing Chrome session through a browser extension. It can collect console output, uncaught exceptions, XHR and fetch activity, failed requests, page details, selected DOM elements, screenshots, browser storage, and connection status. It also exposes tab discovery, page reading, navigation, refresh, limited interaction, and Lighthouse audits for accessibility, performance, SEO, and best practices.

The main distinction is that it observes the Chrome session already open on the developer’s machine. This makes it suitable for investigating an application with an existing login or browsing state, rather than creating a fresh automated profile for testing.

How it works

The setup has one MCP process and a separately loaded Chrome extension. The MCP client starts the package with npx, while the extension connects the browser to the local connector. Capture starts when Chrome DevTools is open for the page being inspected; the BrowserTools panel is used for status and settings.

DevTools-connected tabs use tabId values returned by listBrowserTabs. Tools based on the Chrome tabs API use chromeTabId values returned by listTabs; these identifiers are not interchangeable. Several DevTools tabs can be tracked independently, and results identify their source tab and URL. Most operations use the current inspected tab unless a specific tab ID or allTabs option is supplied.

Large histories are available through MCP resources instead of always being embedded in tool responses. Console and network reads support paging and filters, while screenshots and full Lighthouse reports can be retrieved from linked resources. Read operations are marked read-only in the MCP metadata.

Setup and configuration

The MCP client configuration starts the latest package with:

config.json
{
  "mcpServers": {
    "browser-tools": {
      "command": "npx",
      "args": ["-y", "@agentdeskai/browser-tools-mcp@latest"]
    }
  }
}

Node.js 22.19 or newer is required. On Windows, clients that cannot locate npx can invoke it through cmd /c. The extension is installed by loading the repository’s chrome-extension directory as an unpacked extension from chrome://extensions with Developer mode enabled.

The --doctor command checks the installation. --verbose prints connection and capture activity. The extension popup switch must be enabled for tab operations. Navigation and soft input additionally require --allow-input on the MCP process. Soft actions may return worked or needs a real gesture, rather than asserting that an interaction succeeded.

Tools and capabilities

  • Read console entries with keyword filters, paging, and a separate error-focused view.
  • Inspect network requests, response status, timing, bodies, failed requests, and HAR resources.
  • Identify the current page, selected element, connected tabs, capture counts, and tab contents.
  • Navigate or refresh tabs, take image screenshots, read storage, and clear captured logs.
  • Perform limited clicks and typing through Chrome scripting without attaching the debugger.
  • Run Lighthouse accessibility, performance, SEO, and best-practices audits.

Limitations and notes

The extension captures data visible to the browser, so logs, screenshots, network traffic, and storage require care. The connector binds to 127.0.0.1, and the extension communicates through loopback. Storage values are gated, and credentials are scrubbed before browser data leaves the extension.

listTabs does not represent every possible Chrome tab: incognito tabs are skipped, and some special, pinned, split-view, sleeping, or alternate-space tabs may be absent. Background navigation is limited to the visible window. Sleeping tabs are reported but not reloaded. Soft input is constrained by browser origin rules and may require a real user gesture.

The 1.x line is unsupported and has known security vulnerabilities; use the 2.x rewrite instead.

Reviews

No reviews yet — be the first to share how it worked for you.

Frequently Asked Questions about Browser Tools MCP

browser-tools-mcp is an MCP server that connects MCP clients to an existing Chrome session through a DevTools extension. Its main tools read console and network logs, inspect tabs and pages, capture screenshots, navigate tabs, read storage, clear telemetry, and run Lighthouse accessibility, performance, SEO, and best-practices audits.

Install it by configuring your MCP client to run `npx -y @agentdeskai/browser-tools-mcp@latest`, then load the repository’s `chrome-extension` directory as an unpacked Chrome extension. Node.js 22.19 or newer is required, and DevTools must be open on the page you want to capture.

Yes, browser-tools-mcp works with Cursor. Configure Cursor to launch the package through `npx`, install the Chrome extension, enable its popup switch, and open DevTools on the target page before asking the agent to inspect browser data.

Use `getNetworkErrors` to read failed and 4xx/5xx requests, or use `getNetworkLogs` for broader XHR and fetch activity including status, timing, and bodies. Both log views support bounded reads, and network data can also be retrieved through linked HAR resources.

Yes, within defined limits. It can list tabs, read pages, navigate, refresh, and perform soft clicks or typing, but input may require `--allow-input` and a real gesture. `listTabs` covers only tabs returned by the Chrome tabs API, while DevTools-connected tabs use a separate ID space.

Related MCP servers

Add a badge to your READMEShow it’s listed — live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:
Live Dynamic SVG PreviewBrowser Tools MCP AllMCPs Directory Badge
Markdown (GitHub README)
[![AllMCPs](https://allmcps.com/api/badge/browser-tools-mcp?style=directory)](https://allmcps.com/mcp/browser-tools-mcp)
HTML Embed
<a href="https://allmcps.com/mcp/browser-tools-mcp"><img src="https://allmcps.com/api/badge/browser-tools-mcp?style=directory" alt="Browser Tools MCP on AllMCPs" /></a>