One HTTP call turns any URL into a structured map — headings, sections, forms, links, and metadata. No browser. No JavaScript. Under 2 seconds.
No rendering engine, no DOM traversal, no style computation. Just raw structure — served as clean, readable text.
Full H1→H4 tree in DOM order. Know exactly how a page is organized — sections, subsections, and their nesting.
Identifies every Elementor widget type per section — headings, icon-boxes, text-editors, forms, spacers, and more.
All form fields extracted with placeholders, types, and required flags. Know what data a page expects before you scrape.
Every link on the page with text + href. Navigation menus, footer links, CTAs — all captured in one pass.
All houzez-iconfont and CSS icon classes used on the page. Know which icons a design depends on without inspecting.
Background colors, inline styles, data-settings, CSS classes, and widget types for every section on the page.
No browser, no JavaScript execution, no headless Chrome. Just raw HTTP + pattern matching.
Clean, readable structure — not raw HTML or JSON blob. Designed for human review and agent consumption.
Zero browser dependencies. Pure Node.js HTTP + pattern matching. Runs as a stdio MCP server.
The style-inliner gives you pixel-perfect CSS. This one gives you the blueprint.
Before building a page, understand its section structure, heading hierarchy, and form fields.
Map out every heading, link, and form on a page in seconds — no manual scrolling.
Identify Elementor widgets, section settings, and icon dependencies without logging into WP admin.
Before an agent edits a page, load its structure so it knows exactly what exists and where.
Use together for full page replication: structure first, then styles.
| Feature | page-structure MCP | style-inliner MCP |
|---|---|---|
| Speed | ~1-2s | ~1-2s (90 properties) |
| Browser required | ❌ No | ✅ Headless Chrome |
| CSS computed | ❌ | ✅ Full inline |
| Heading tree | ✅ Full hierarchy | ❌ |
| Form fields | ✅ With placeholders | ❌ |
| Max page size | 2MB | ~500KB (before timeout) |
| Use case | Blueprint / structure | Pixel-perfect export |
Already registered in this Hermes session. Just reload MCP and call extract_structure(url)