MCP server for Chrome DevTools-based UI testing and automation, providing tools for page navigation, DOM inspection, script evaluation, and screenshot capture
14 tools with basic schemas and descriptions, but significant gaps in parameter documentation and error handling. Most tools lack descriptions for optional parameters (targetId, pageId appear in 8 tools with identical descriptions repeated). No output schemas documented. Parameter descriptions are minimal (avg ~40 chars). No error recovery guidance. Naming is verb-forward (good), but composition issues exist: select_page, navigate_page, evaluate_script, click, take_screenshot, take_snapshot all accept redundant targetId/pageId alternatives, forcing LLMs to reason about parameter equivalence. No idempotency hints. Security: evaluate_script is high-risk (arbitrary JS execution) with no warnings or sandboxing notes.
Click an element on the page by selector or uid
Detect UI component library in use (Arco Design, Ant Design, Element UI, Element Plus, MUI, etc.)
Collect all form elements on the page with their types, selectors, current values, and options
Detect the JavaScript framework used on the page (Vue2, Vue3, React, Angular, jQuery, etc.)
Identify page structure patterns (list+detail, wizard, dashboard, form page, etc.)
Evaluate JavaScript code in the page context and return the result
Vue 2 specific: recursively find components holding form data and list their methods
Check the health status of the CDP proxy and Chrome connection
Redundant parameter alternatives (targetId vs pageId) repeated across 8 tools force LLMs to reason about equivalence. No guidance on which to prefer or when they conflict.
evaluate_script accepts arbitrary JavaScript with no sandboxing, timeout, or error recovery guidance. High security risk and potential for runaway execution.
No output schemas documented for any tool. LLMs cannot predict response structure, forcing them to guess field names for downstream tool calls.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 50 | 2026-07-28+ | v2 |
List all open pages/tabs in the Chrome browser
Navigate a page to a URL or perform navigation actions
Create a new page/tab and optionally navigate to a URL
Select a specific page/tab by targetId or pageId
Take a screenshot of the page and optionally save to file
Take a DOM snapshot of the current page, returning both text and structured element list
Parameter descriptions for targetId and pageId are identical across tools ('Chrome target ID' / 'Zero-based page index') but lack guidance on mutual exclusivity or precedence.
No error handling guidance. Tools do not document what happens on failure (e.g., invalid selector, page not found, script timeout) or how to recover.