Read Axe accessibility reports with AI
Keep an issue attached to the right element when moving from an Axe scan to a proposed fix, including findings from guided tests.
Teams that read Axe accessibility reports with AI need to know what a repeated rule actually represents. The mcp-usage skill explains how a connected agent should interpret scan output, carry the right evidence into remediation and decide which checks are worth repeating.
A grid of product cards might contain dozens of instances of one rule. A shared button component may need one source edit, while ten images of different products need separate descriptions. You can ask for a smaller scope before spending credits on suggested fixes.
What it does
- Match the element, rule and guidance across the scan and remediation results.
- Keep guided keyboard findings distinct from ordinary automated checks.
- Review incomplete results when a focus trap prevented the rest of a test.
- Narrow a large set of findings to one component or impact level.
- Run a fresh ordinary scan after editing and request a paid guided recheck separately.
Install and use
- Install the skill, connect the official Axe tools and use a permitted test page without personal data. Paid remediation and guided tests consume credits.
- Ask “Show me the serious issues in the filter panel, with their affected controls. Tell me how many before asking for fixes.”
- Compare each suggestion with the actual component, then rescan the agreed area.
FAQ
Source
Made by Deque Systems. Original skill: mcp-usage, published on GitHub under the MIT licence.
View on GitHub