All products

Review alt text with AI

Inspect accessible names in real interface code. Distinguish vague writing from a template variable that was never resolved, and get a specific fix for each.

To review alt text with AI, give your agent a component that renders a real image or control. A product photo can look entirely normal while its accessible name is the literal text of a template variable. The text-quality-reviewer skill helps the agent trace where the label comes from and decide whether the wording or the binding needs to change.

The distinction affects the fix. Rewriting “product photo” once in a template will give every item the same description. Connecting the image to the correct product field may solve the underlying problem. Where an image communicates a detail that the data does not describe, supply the image or a reliable account of what it shows.

What it does

  • Find template placeholders that could be spoken to a visitor.
  • Inspect a button whose visible and accessible names suggest different actions.
  • Identify decorative images that add unnecessary narration.
  • Check the fallback text shown when a product lacks a description.
  • Separate a missing content field from an incorrectly written framework binding.

Install and use

  1. Install the skill in your coding agent. It needs the relevant markup or component and no separate service.
  2. Explain the purpose of the control or image and supply any visual context needed.
  3. Review each finding in the rendered interface using representative content.

FAQ

Source

Made by Community Access (Taylor Arndt). Original skill: text-quality-reviewer, published on GitHub under the MIT licence.

View on GitHub