Review image alt text with AI
Read an image description in its page context, then inspect the headings and page title that help someone find it. Capture precise issues for an editor.
Review image alt text with AI by considering why the editor chose the image. A photograph of a brown boot on a product page describes the item for sale. The same photograph in a repair guide might show a split seam. Its colour is still visible, but the damaged area now carries the information a reader needs.
The alt-text-headings specialist inspects images, their surrounding content and the page outline. It can distinguish a decorative illustration from a meaningful image or from an icon that performs an action. Give it the image itself whenever the accuracy of a description is in question.
What it does
- Compare a description with the image in context
- Identify a decorative image that adds reading noise
- Check the accessible name of an icon acting as a button
- Follow the heading hierarchy through a long page
- Distinguish pages with meaningful titles
Install and use
- Use a compatible agent with the original specialist references and shared findings schema. Supply authorised page markup and images. Visual assessment requires a host that can inspect the actual image.
- Ask, “This photo explains where a seam has split. Check whether the current description conveys that detail, and flag anything you cannot establish from the image.”
- Listen to the result in context or read the surrounding paragraph and caption aloud. Also try navigating by headings to the relevant repair step.
FAQ
Source
Made by Community Access. Original skill: alt-text-headings, published on GitHub under the MIT licence.
View on GitHub