Opens in a new tabSkip to content
Agent LighthouseAgent Lighthouse

    Searches the text of every published page. The evidence sources themselves are not in this index — search all of them on the trusted sources page.

    GitHub ↗
    Browse checks and page contents
    content-extraction/figure-figcaption

    <figure> + <figcaption> usage

    What it checks

    AI agents use <figcaption> to understand the purpose and context of figures beyond what alt text provides. Without captions, agents treat figures as opaque image containers with no semantic meaning, missing opportunities to cite your visual data in AI-generated answers.

    Why it matters

    Wrapping an image in <figure> with a <figcaption> creates a programmatic association between the caption text and the image. The caption is then exposed as the figure’s accessible name or description, and it stays adjacent to the image reference through extraction and markdown conversion. Without it, the caption floats as an unattached paragraph whose relationship to the image must be inferred from proximity.

    Evidence

    figure

    The mapping is specified: HTML-AAM maps figure→figure role and figcaption→caption role [w3c-html-aam], and accname’s recursive name-from-content rules let figcaption text contribute to the figure’s accessible name [w3c-accname]. Google states it draws image context from ‘captions and image titles’ alongside alt text and surrounding copy [google-image-seo-docs]. Playwright and Chrome DevTools snapshots surface figure nodes like any other role [playwright-mcp-snapshots, chrome-devtools-mcp-tool-reference].

    Limits

    Weak and largely inferential. No AI vendor names figcaption anywhere. Mozilla Readability does not preserve or prioritise figcaption — figures are touched only by the pass that recovers lazy-loaded image sources [mozilla-readability-source] — and trafilatura’s documented preservation list covers tables, lists, headings and formatting without singling out figcaption [trafilatura-corefunctions]. Google’s caption statement does not distinguish <figcaption> from a nearby <p>, so the specific claim that the ELEMENT (rather than mere text proximity) is what helps is unproven. Adoption is not measured in the Web Almanac accessibility chapter. This is a plausible mechanism with no demonstrated consumer of the binding itself — informative, never scored.

    Sources