Inspect heading levels without rendering the page
Paste HTML and choose Inspect HTML to list H1 through H6 elements in parsed document order. Each row contains the level, whitespace-collapsed descendant text and id attribute. This helps review source structure without loading the original site's images, styles or scripts.
Example: a level jump and a hidden heading
The example contains H1 Guide with id main, H3 Setup notes, and H2 Hidden section. All three appear in that order. The hidden attribute does not remove the third row because this is a source inspection, not a visible-content calculation. The H1-to-H3 jump is shown as written; no invented SEO score is assigned.
What an outline can and cannot establish
Only actual H1-H6 tags are included. A styled div, an ARIA heading role, or text that looks large in a screenshot is not inferred as a heading. The tool does not repair level order, generate anchor IDs or remove repeated headings. IDs are shown literally and are not checked for uniqueness across an external website.
Descendant text is not the computed accessible name. CSS visibility, generated content, script-inserted headings and rendering effects are outside the report. HTML parsing may repair malformed nesting. Headings inside nested template contents are not traversed. Keep the source if exact formatting or byte offsets matter; row order is not a source line number.
Export a structural snapshot
Parsing uses a detached template that is never inserted into the page. The browser's template fragment model is different from rendering a live document. Scripts are not executed and referenced resources are not fetched.
The first 200 rows appear in the table; JSON copy and download include all rows. Limits are 100,000 UTF-16 input units, 10,000 parsed elements and 2,000 report rows, with a conservative two-million-character report budget. No matching headings is a valid empty result. Changing the source clears previous results. Use HTML Link Inspector for anchor attributes or Markdown Heading Extractor when the input is Markdown rather than HTML.





