Information Architecture Audit: AI Skill for UX Design
Information Architecture Audit is an AI skill that reviews sitemaps, navigation trees, and category structures. It tests labels for information scent, checks hierarchy depth, and rates every finding by how badly it strands users.
npx skills add Uxcel-Lab/product-skills --skill ux-information-architecture-audit
What this skill does
Turns eyeball critiques of site structure into a principle-based IA audit with severity-rated fixes.
- Checks seven IA areas, from organization schemes to progressive disclosure.
- Tests every label for information scent: could a user predict the destination?
- Flags narrow-and-deep hierarchies, lopsided categories, and structures that fail the growth test.
- Walks in from deep links and search, not just the homepage-down path.
- Checks findability across seeking modes: search, browsing, and re-finding.
- Names the research method that settles contested calls: card sorts, search logs, free listing.
- Rates every finding by lostness cost, then names the top three fixes.
- Runs standalone or as a validation step after navigation or menus get generated.
When to use it
Use it when you need an AI assistant to:
- audit a sitemap, navigation tree, or category structure;
- test menu labels for scent, consistency, and vocabulary fit;
- check whether a hierarchy is too deep or lopsided;
- find dead ends and orphan pages before users do;
- confirm a large catalog supports search, filters, and re-finding;
- validate AI-generated navigation before it ships.
What this audit checks
| Area | What it catches |
|---|---|
| Organization scheme | A-to-Z ordering where users browse by topic, schemes that fight top user tasks |
| Structure & depth | Content buried four levels deep, lopsided categories, structures that fail the growth test |
| Labeling system | CTA verbs as category names, mixed syntax and granularity, audience-mismatched vocabulary |
| Information scent | Generic "Learn more" links, labels and imagery that mispredict the destination |
| Findability | Browse-only large catalogs, missing search, nothing that supports re-finding |
| Every page a front door | Orphan pages, dead ends, deep pages with no orientation signals |
| Choices & disclosure | Mega-menus exposing everything at once, key tasks buried in exhaustive lists |
How the skill works
1
Establish context
The skill confirms content volume, top user tasks, and the seeking modes the structure must serve.
2
Walk the entry points
It audits as a deep-link arrival, a known-item searcher, and a browser, not just homepage-down.
3
Report by lostness cost
It rates each finding, gives the fix, and names the research method that settles contested calls.
Try it with
Related skills
1.0.0
MIT
UX Design
Audits
