Fox Chrome Devtools Mcp

Chrome DevTools MCP — Google's official browser automation and testing server. Control Chrome via Puppeteer through MCP protocol: click, fill forms, navigate...

Security Scan

The skill's requirements, instructions, and included setup script are consistent with a Chrome DevTools MCP integration and do not request unexplained credentials or access.

Purpose & Capability

Name/description (MCP control of Chrome) match the instructions and included script: Node/npm usage, optional Chrome/Chromium, and an OpenClaw MCP server config. No unrelated credentials, binaries, or config paths are requested.

Instruction Scope

SKILL.md only instructs installing/running the chrome-devtools-mcp npm package via npx and adding an MCP entry to openclaw.json. The supplied Python script performs expected checks (node, npx, Chrome paths, openclaw.json) and offers a test start; it does not read or transmit unrelated secrets or contact unknown endpoints beyond npm/npx/Chrome (expected for this tool).

Install Mechanism

There is no formal install spec in the package; instructions use npx -y chrome-devtools-mcp@latest which will fetch and execute code from the npm registry at runtime. That is expected for this kind of integration but represents the usual supply-chain risk of running code fetched from npm and of using the unpinned "latest" tag.

Credentials

The skill declares no required environment variables, no credentials, and checks only reasonable local paths (Chrome and a Playwright cache path). It does read ~/.openclaw/openclaw.json to look for existing config, which is relevant to its purpose.

Persistence & Privilege

The skill does not request always:true, does not modify other skills, and only suggests adding an MCP server entry to the agent config (it prints the config but does not write to global settings). Autonomous invocation is allowed (platform default) but is not combined with other high-risk factors here.

Assessment

This skill appears to be what it claims: an integration for controlling Chrome via the chrome-devtools-mcp npm package. Before installing, consider: (1) npx -y chrome-devtools-mcp@latest will fetch and run code from the npm registry — pin a specific, audited version instead of relying on 'latest' to reduce supply-chain risk; (2) run the MCP server in a sandboxed environment (container/VM) if you'll be browsing untrusted sites or handling sensitive data, since browser sessions can expose form data and credentials; (3) disable telemetry/performance CrUX flags as shown if you don't want traces or usage stats sent externally; and (4) review your openclaw.json after adding the MCP entry. No unexplained credential or file access was found in the provided files.

🌐 Chrome DevTools MCP

Google's official Chrome DevTools MCP server — gives AI agents full control of a live Chrome browser via Puppeteer and the Chrome DevTools Protocol.

Features

• Input automation — click, drag, fill forms, hover, press keys, upload files, handle dialogs
• Navigation — open/close/switch pages, wait for elements/network idle
• Screenshots & snapshots — capture page state visually and as DOM
• Performance traces — record and analyze Chrome performance traces with insights
• Network inspection — list/inspect network requests and responses
• Console debugging — read console messages with source-mapped stack traces
• Device emulation — emulate mobile devices, resize viewport
• Form automation — fill multiple form fields at once

Requirements

• Node.js v20.19+ (already available in OpenClaw)
• Chrome/Chromium browser

Quick Start

Install & verify

npx -y chrome-devtools-mcp@latest --help

Start the MCP server

# Standard (launches Chrome automatically)
npx -y chrome-devtools-mcp@latest

# Headless mode (for servers)
npx -y chrome-devtools-mcp@latest --headless

# Connect to existing Chrome (must be started with --remote-debugging-port=9222)
npx -y chrome-devtools-mcp@latest --browser-url=http://127.0.0.1:9222

# Disable telemetry
npx -y chrome-devtools-mcp@latest --no-usage-statistics --no-performance-crux

OpenClaw MCP Integration

Add to your openclaw.json under MCP servers:

{
  "mcp": {
    "servers": {
      "chrome-devtools": {
        "command": "npx",
        "args": ["-y", "chrome-devtools-mcp@latest", "--headless", "--no-usage-statistics"]
      }
    }
  }
}

Or use the setup script:

python3 {baseDir}/scripts/setup_chrome_mcp.py setup
python3 {baseDir}/scripts/setup_chrome_mcp.py status
python3 {baseDir}/scripts/setup_chrome_mcp.py test

Tool Reference

Input Automation (8 tools)

Tool	Description	Key Params
click	Click an element	uid (required), dblClick
drag	Drag element onto another	from_uid, to_uid
fill	Type text into input/textarea/select	uid, value
fill_form	Fill multiple form elements at once	elements[]
handle_dialog	Accept/dismiss browser dialogs	action (accept/dismiss)
hover	Hover over element	uid
press_key	Press keyboard key	key
upload_file	Upload file to input	uid, paths[]

Navigation (6 tools)

Tool	Description	Key Params
navigate_page	Go to URL	url
new_page	Open new tab	url
close_page	Close current tab	—
list_pages	List all open tabs	—
select_page	Switch to tab	index
wait_for	Wait for element/network	event, uid, timeout

Debugging (5 tools)

Tool	Description
take_screenshot	Capture page as image
take_snapshot	Get DOM/accessibility snapshot
evaluate_script	Run JavaScript in page
list_console_messages	Get console log entries
get_console_message	Get specific console message

Performance (3 tools)

Tool	Description
performance_start_trace	Begin performance recording
performance_stop_trace	Stop and get trace data
performance_analyze_insight	AI analysis of trace

Network (2 tools)

Tool	Description
list_network_requests	List all network requests
get_network_request	Get request/response details

Emulation (2 tools)

Tool	Description
emulate	Emulate device (mobile, tablet)
resize_page	Change viewport size

Common Workflows

Test a webpage

1. navigate_page → URL
2. take_snapshot → get element UIDs
3. click/fill → interact with elements
4. take_screenshot → capture result

Performance audit

1. navigate_page → URL
2. performance_start_trace
3. Interact with page
4. performance_stop_trace
5. performance_analyze_insight

Form testing

1. navigate_page → form URL
2. take_snapshot → identify form fields
3. fill_form → fill all fields at once
4. click → submit button
5. take_screenshot → verify result

Privacy Notes

• Google collects usage statistics by default — disable with --no-usage-statistics
• Performance tools may send trace URLs to Google CrUX API — disable with --no-performance-crux
• Avoid sharing sensitive data in browser sessions