chrome-devtools

krazyuniks's avatarfrom krazyuniks

Chrome DevTools MCP for interactive debugging and development. Use for CSS inspection, network debugging, console access, performance profiling, and real-time development iteration.

⭐0stars🔀0forks📁View on GitHub🕐Updated Jan 8, 2026

When & Why to Use This Skill

The Chrome DevTools MCP skill empowers Claude to perform interactive, real-time web debugging and development by connecting directly to a running Chrome browser. It bridges the gap between code and execution, allowing for live CSS inspection, network traffic analysis, console monitoring, and performance profiling to drastically shorten frontend development cycles.

Use Cases

  • CSS & Layout Troubleshooting: Inspect computed styles, box model properties, and element dimensions to quickly identify and fix layout shifts or CSS specificity issues.
  • Network Request Debugging: Monitor real-time API calls to inspect request/response headers, payloads, and status codes, helping resolve integration errors and data mismatches.
  • Live Console Monitoring: Capture JavaScript errors and logs as they happen during development, enabling immediate identification and resolution of runtime bugs.
  • Performance Profiling: Access detailed metrics for CPU usage, memory consumption, and paint timings to optimize web application speed and responsiveness.
  • Interactive State Exploration: Navigate pages, click elements, and execute custom JavaScript within the browser context to verify application behavior and state changes.
namechrome-devtools
descriptionChrome DevTools MCP for interactive debugging and development. Use for CSS inspection, network debugging, console access, performance profiling, and real-time development iteration.

Chrome DevTools Skill

Activation: debugging, CSS inspection, network debugging, console, performance, real-time development, layout issues, computed styles

Overview

Chrome DevTools MCP connects to your running Chrome browser for interactive debugging. Use it during development for faster iteration cycles.

Key Difference from Playwright:

  • Chrome DevTools = Interactive development, debugging, exploration
  • Playwright = Automated testing, screenshot verification, CI/CD

When to Use Each

Task Chrome DevTools Playwright
CSS debugging Yes - Live inspect, edit styles No
Console errors during dev Yes - Real-time monitoring Yes - Snapshot
Network inspection Yes - Full request/response details Yes - Summary
Performance profiling Yes - CPU, memory, rendering No
Screenshot proof for PR No Yes - Required
Automated E2E tests No Yes - Primary tool
Element state inspection Yes - Computed styles, box model Yes - Accessibility tree
Storage inspection Yes - localStorage, cookies, IndexedDB No

MCP Tools

Page Interaction

mcp__chrome-devtools__navigate - Navigate to URL
mcp__chrome-devtools__screenshot - Capture screenshot
mcp__chrome-devtools__click - Click element
mcp__chrome-devtools__type - Type text
mcp__chrome-devtools__evaluate - Run JavaScript in page context

Debugging

mcp__chrome-devtools__get_console_logs - Get console output
mcp__chrome-devtools__get_network_logs - Get network requests with full details
mcp__chrome-devtools__get_computed_style - Get computed CSS for element
mcp__chrome-devtools__get_element_info - Get element properties, box model

Performance

mcp__chrome-devtools__get_performance_metrics - CPU, memory, paint timing
mcp__chrome-devtools__start_profiling - Begin CPU profile
mcp__chrome-devtools__stop_profiling - End and get profile data

Development Workflow

1. CSS Debugging

When something doesn't look right:

1. Get element info
   → mcp__chrome-devtools__get_element_info selector=".my-element"
   → Shows: box model, dimensions, position

2. Get computed styles
   → mcp__chrome-devtools__get_computed_style selector=".my-element"
   → Shows: All applied CSS, computed values

3. Identify the issue (wrong padding, unexpected margin, etc.)

4. Fix in code, hot reload, verify

2. Network Debugging

When API calls aren't working:

1. Get network logs
   → mcp__chrome-devtools__get_network_logs
   → Shows: All requests with status, timing, headers

2. Find failing request

3. Get request details
   → Full request/response headers
   → Request body
   → Response body (if error)

4. Fix issue (wrong endpoint, missing auth, etc.)

3. Console Monitoring

During development iteration:

1. Make code change

2. Check console immediately
   → mcp__chrome-devtools__get_console_logs
   → Catch errors as they happen

3. Fix and iterate quickly

4. Layout Investigation

When layout breaks:

1. Get element info for parent
   → Check dimensions, display mode

2. Get computed styles
   → Check flex/grid properties
   → Check overflow behavior

3. Get child element info
   → Compare expected vs actual sizes

4. Identify constraint causing issue

Complementary Use with Playwright

Development Phase (Chrome DevTools)

You: "The button styling looks wrong"

1. Navigate to page (DevTools)
   → mcp__chrome-devtools__navigate url="http://localhost:3000/page"

2. Inspect button styles
   → mcp__chrome-devtools__get_computed_style selector="button.primary"

3. Find issue: wrong color value applied

4. Fix CSS in code

5. Verify fix visually (DevTools)
   → mcp__chrome-devtools__get_computed_style (confirm new value)

PR Phase (Playwright)

Once feature is working:

1. Run through screenshot-eval checklist
   → mcp__playwright__browser_navigate
   → mcp__playwright__browser_console_messages
   → mcp__playwright__browser_network_requests
   → mcp__playwright__browser_take_screenshot

2. Include screenshot proof in PR

Common Use Cases

"Why isn't my CSS applying?"

1. Get computed style
   → mcp__chrome-devtools__get_computed_style selector=".my-class"

2. Look for:
   - Value different from expected → specificity issue
   - Value shows "inherit" or "initial" → rule not matching
   - Value correct but element wrong → targeting issue

"My API call returns wrong data"

1. Get network logs
   → mcp__chrome-devtools__get_network_logs

2. Find the API call, check:
   - Request URL correct?
   - Request headers (auth token)?
   - Request body (if POST)?
   - Response status code?
   - Response body contents?

"Something is slow"

1. Get performance metrics
   → mcp__chrome-devtools__get_performance_metrics

2. Look for:
   - High CPU usage → expensive computation
   - Memory growing → leak
   - Long paint times → complex rendering

"Element not where expected"

1. Get element info
   → mcp__chrome-devtools__get_element_info selector=".element"

2. Check box model:
   - Content size
   - Padding
   - Border
   - Margin

3. Compare with parent container
   → mcp__chrome-devtools__get_element_info selector=".parent"

Setup

Chrome DevTools MCP requires:

  1. Chrome browser running with remote debugging enabled
  2. MCP server configured in Claude Code settings

The agent will connect to your existing Chrome session, so you can see changes in real-time.

  • playwright - For automated testing and PR verification
  • screenshot-eval - For systematic error checking
  • frontend-dev - For Astro + React patterns
  • frontend-design - For design principles