What is the UI audit skill?
The UI audit skill sits in the sundial-org/awesome-openclaw-skills collection and is based on Making UX Decisions by Tommy Geoco. It gives an AI coding agent reference files on hierarchy, style, accessibility, navigation and forms, then asks it to produce a JSON audit with pass, warn or fail checks and ranked fixes.
It is the ui-audit skill in sundial-org/awesome-openclaw-skills. Its description calls it an AI skill for automated UI audits that checks interfaces against UX principles, and says it is based on Making UX Decisions by Tommy Geoco.
The skill folder holds a SKILL.md plus a references folder of numbered files: a core framework, anchors, three checklists for new interfaces, fidelity and visual style, an innovation checklist, and twelve pattern files.
What changes in your agent's output
- Design decisions follow a three step flow: weigh information from institutional knowledge, user familiarity and research, narrow options against constraints, then execute with a checklist.
- Audit requests return a structured JSON report with a title, macro bets, jobs to be done, sections of checks and a ranked priority fix list.
- Each check carries a status of pass, warn, fail or na, with notes, and each priority fix can cite a framework reference file.
- Visual hierarchy, visual style and accessibility are always included, while navigation, usability, onboarding, social proof and forms appear when relevant.
Who the UI audit skill is for
Best for
- Designers and developers who need a repeatable review of a screen before build or release.
- Teams making design trade-offs under time pressure who want the reasoning tied to business context.
- Pairing with the design review skill or the UX heuristics skill for a second angle.
Not for
- Automated pixel or code scanning. The skill is a reasoning framework and report format, not a scanner.
- Users who need a licensed redistributable package, since the repository itself shows no license.
Use it with UiChemy
Audit a landing page or Figma frame before you convert it. Fix hierarchy and accessibility findings in the design, then use UiChemy Figma to WordPress to convert it. Click Get Started on uichemy.com. For component states, see the critique affordance skill.
How to install UI Audit Skill: UX Decision Framework for AI Agents
npx sundial-hub add ui-auditThe skill folder also has a CLAUDE.md and a short README. The repository root has no license file.Use the collection installer
The repository README lists
npx sundial-hub add <name>as its install command. Replace the name with ui-audit.Or copy the folder
Clone the repository with git and copy skills/ui-audit, including its references folder, into your agent's skills directory.
Keep the references
SKILL.md points to numbered files in references. The agent loads them as needed, so copy the whole folder.
Ask for an audit
Share a screen, URL or description and ask for a UI audit. Mention the product goals so macro bets can be filled in.
Check that it loaded
Ask for an audit of one screen. The agent should answer with the JSON structure, marked checks for hierarchy, style and accessibility, and a ranked list of priority fixes.
Key features
- Three step decision flowWeigh institutional knowledge, user familiarity and research, then narrow options and execute with checklists and patterns.
- Three pillarsScaffolding for recurring decisions, decisioning for new ones and crafting through checklists.
- Macro betsVelocity, efficiency, accuracy and innovation, each with a design implication, so choices align with company goals.
- Four checklistsNew interfaces in six steps, fidelity, visual style covering spacing, color, elevation, typography and motion, and an innovation spectrum.
- Twelve pattern filesChunking, progressive disclosure, cognitive load, visual hierarchy, social proof, feedback, error handling, accessibility, personalization, onboarding, information and navigation.
- JSON audit formatA defined report shape with macro bets, jobs to be done, check lists per section and numbered priority fixes.
- Required and contextual sectionsHierarchy, style and accessibility always appear, others only when the screen calls for them.
- Check statusesEvery check is marked pass, warn, fail or na, with six to ten checks suggested per section.
- Accessibility checksKeyboard operability, visible focus, 4.5 to 1 contrast, 44px touch targets, alt text, semantic markup and reduced motion support.
- Reference links in fixesPriority fixes can point to a numbered reference file and section, so each recommendation is traceable.
Frequently asked questions
What is the UI audit skill?
It is a skill in the sundial-org/awesome-openclaw-skills collection that gives an AI coding agent a UX decision framework and reference files, then has it produce a structured audit of an interface with checks and ranked fixes.
Who wrote the framework behind it?
The skill description and homepage field credit Tommy Geoco and say it is based on Making UX Decisions. The collection repository lists it among its skills, so check the skill folder for authorship details.
How do I install the UI audit skill?
Run the collection's install command, npx sundial-hub add ui-audit, or clone the repository and copy skills/ui-audit with its references folder into your skills directory. The README documents the first route.
What does the audit report look like?
It is JSON with a title, project, date, macro bets, jobs to be done, check sections and priority fixes. Each check has a label, a pass, warn, fail or na status and notes.
Does it scan my code automatically?
No. The skill is a set of instructions and reference files that guide the agent's judgment and report format. It does not describe a crawler or scanner, so you supply the design or page to review.
Does it cover accessibility?
Yes. Accessibility is a required section, covering keyboard operability, focus, color contrast, touch targets, alt text, semantic markup and reduced motion. A dedicated accessibility patterns reference file backs it, and a WCAG audit skill goes deeper.
What license applies?
The repository has no license file, and GitHub shows none. The skill's own package.json lists MIT, but treat the licensing as unclear for the whole collection and check with the author before you redistribute the files.


