browser_mcp
Headless Chrome automation. Supported agents can navigate pages, take screenshots, click elements, fill forms, run JavaScript, and scrape content — useful for testing, web research, and debugging production issues.
Tools
| Tool | Parameters | What it does |
|---|---|---|
navigate | url | Navigate to a URL |
screenshot | full_page? | Capture a screenshot — returns it inline in chat |
get_page_content | selector? | Get page text content, optionally scoped to a CSS selector |
get_page_html | selector? | Get raw HTML, optionally scoped to a CSS selector |
click | selector | Click an element |
type_text | text, selector?, clear_first? | Type text into a field |
evaluate_js | expression | Run arbitrary JavaScript in the page context |
get_console_logs | clear? | Get browser console output |
wait_for_selector | selector, timeout_ms? | Wait for an element to appear in the DOM |
close_browser | — | Close the browser session |
Chrome launches automatically on first use and stays running for the duration of the session. You don't need to install anything — it's bundled with the flamel Docker image.