Chrome DevTools MCP Testing

Goal: quickly verify user flows and collect actionable evidence (console, network, screenshots, traces).

Prerequisites (no hardcoded secrets)

• Read test env config from environment variables when available:
  • TEST_BASE_URL
  • TEST_EMAIL
  • TEST_PASSWORD
• If any are missing, ask the user (do not invent credentials and do not write them into files).

Workflow

1) Navigate

• Open a page with mcp__chrome-devtools__new_page (or ...__navigate_page for an existing page).
• Prefer going to the login page directly if auth is required.

2) Login (if needed)

• Use mcp__chrome-devtools__take_snapshot to locate form inputs and buttons.
• Use mcp__chrome-devtools__fill / ...__fill_form and mcp__chrome-devtools__click.
• Use mcp__chrome-devtools__wait_for for a post-login indicator (URL change, dashboard text, etc.).

3) Execute the test scenario

• Interact with the UI via ...__click, ...__fill, ...__press_key.
• If visual verification helps, use mcp__chrome-devtools__take_screenshot.
• To check errors, use mcp__chrome-devtools__list_console_messages.
• To inspect API calls, use mcp__chrome-devtools__list_network_requests + ...__get_network_request.

4) Performance (optional)

• Record a trace with mcp__chrome-devtools__performance_start_trace and ...__performance_stop_trace.
• Summarize CWV/insights relevant to the scenario.

Safety / Reporting

• Never log or store passwords/tokens in plaintext.
• If a CAPTCHA/Turnstile appears, prompt the user to intervene manually.
• Report: steps executed, pass/fail, console errors, failing network requests, and attached screenshots/traces.