Review interface accessibility with an agent
Turn scoped visual and interaction checks into an evidence-based report, colour proposals and a practical correction plan.
Review interface accessibility with an agent to identify readable colour combinations, visible controls and specific changes worth making. The skill works from authorised source, designs, screenshots or an available running interface. It separates a product review from changes to your own editor or terminal appearance.
Its report keeps observed problems, measured results and untested behaviour distinct. Ask, for example, for a booking form review that preserves your brand colours. You receive a prioritised action plan and useful colour choices, with the evidence needed to judge each finding. A visual mock-up remains a proposal until the relevant behaviour is checked in the real interface.
What it does
- Compare text and background combinations using measured contrast
- Review form errors, control states and visible keyboard focus
- Propose a small palette with explicit functional roles
- Find text scaling and layout problems within available coverage
- Improve supported personal editor or terminal appearance settings
- Track corrections alongside checks that remain pending
Install and use
- Supply the complete folder to a compatible agent and invoke the skill explicitly. For direct instruction use, provide SKILL.md and GUIDE.md; a repository link alone does not load them.
- Identify the target, authorised scope, fixed brand assets and available evidence. State whether you want a review or implementation of agreed changes.
- Compare the findings and colour proposals. Each problem should include evidence, a suggested correction and a verification step; missing tools must leave a visible gap.
- Apply authorised changes and recheck affected states. Record previous values and a restoration path before changing personal settings.
FAQ
Source
Made by antorome. Original skill: accessibility, published on GitHub under the MIT licence.
View on GitHub