skill
ia-practitioner
A Claude skill for professional IA — built from 7 practitioner books including Spencer, Morville & Rosenfeld, Krug, and Marquis.
Set up
Skills are installed by placing their folder where your agent loads skills.
The download comes from github.com/sidhanth-povil/ia-practitioner itself, not from our server, so what you get is exactly what its author published.
Install instructions come from the README
README
synced Oct 17, 2026ia-practitioner
A Claude skill for professional Information Architecture
Apply practitioner IA frameworks to any website, app, or content project — directly in conversation.
What this skill does
Install once and Claude becomes a working IA consultant. It applies a rigorous framework built from seven practitioner texts — covering the full lifecycle from user research through navigation design, search systems, taxonomy, testing, and client deliverables.
| Capability | What Claude does |
|---|---|
| Nav audits | Runs the full six-stage audit methodology — framing, structural walk, ten evaluation lenses, issue cards with severity and effort ratings, restructure proposal, nine-section client report |
| Classification | Identifies which of the 8 classification schemes and 10 IA patterns best fit your content and users |
| Search systems | Designs search architecture — zones, result presentation, autocomplete, faceted navigation, no-dead-ends policy |
| Vocabulary control | Builds controlled vocabularies, synonym rings, authority files, and thesauri — consistent on-site and off-site |
| Research planning | Runs card sort and tree test sessions — writes scenarios, advises on tools and participant numbers, analyses results qualitatively and quantitatively |
| Quantitative analysis | Reads similarity matrices, dendrograms, and standardisation grids; reports tree-test success, directness, and first-click metrics with defensible sample sizes |
| Audit reports | Generates client-ready .docx reports with before/after nav comparisons, issue cards, and roadmaps |
| Labelling | Tests every label against user terminology; catches internal jargon, audience-nav traps, junk drawers |
| Healthcare IA | Applies the symptom → condition → treatment ladder, plain-language labelling, and regulatory vocabulary control for clinic and medical sites |
| Cross-channel IA | Applies Resmini & Rosati's five heuristics for consistent experiences across devices and channels |
| Taxonomy design | Documents controlled vocabularies, tagging systems, and faceted classification for CMS-driven sites |
Install
Method 1 — Claude Skills (global, recommended)
1. Download ia-practitioner.skill (repo root or Releases)
2. Open Claude.ai → Settings → Skills → Install from file
3. Drop the .skill file in
Works immediately across all your Claude conversations. No configuration needed.
Method 2 — Claude Projects (per-project)
1. Create a new Project in Claude.ai
2. Upload SKILL.md and the entire references/ folder into the project Knowledge base
3. Run contextual audits within that project
Best for agency workflows where each client gets its own project context.
Prompt starters
After installing, try any of these:
"Audit the IA of gipartnersofil.com"
"Run a full IA audit of [url] — issue cards and severity ratings"
"What classification scheme suits a portfolio site?"
"How should I structure a healthcare clinic website?"
"Help me write tree testing scenarios for my new nav"
"Analyse these tree test results — is 68% success acceptable?"
"How many participants do I need for a quantitative card sort?"
"What nav pattern suits an e-commerce site with 200 products?"
"Generate an IA audit report for [url] as a Word doc"
"My client's navigation feels wrong — how do I test it?"
"Run an open card sort analysis on these 30 items"
"What's wrong with using an audience-based nav as the primary scheme?"
"Design a search system for a 10,000-page university site"
"Build a controlled vocabulary for a healthcare clinic"
"How do I design faceted navigation for a product catalogue?"
Skill structure
ia-practitioner/
├── SKILL.md ← Core — loaded on every relevant conversation
└── references/
├── classification-schemes.md ← All 8 schemes; LATCH framework; controlled vocabulary; thesauri
├── ia-patterns.md ← 10 IA patterns; cross-channel patterns; Glushko organising systems
├── user-analysis.md ← Research methods; quantitative card sort + tree test analysis; structural audits; verification states
├── navigation-design.md ← Nav types; search systems; wayfinding; faceted nav; healthcare IA; anti-patterns
└── audit-methodology.md ← Six-stage audit process; evaluation lenses; issue cards; severity; client report structure
SKILL.md loads on every relevant conversation — kept tight and scannable. Reference files are loaded on demand when the depth of a question requires them.
Live example
The skill ships with a full audit of gipartnersofil.com — a multi-location GI medical group in Illinois.
Five violations caught immediately:
- Org-chart drift — Locations organised by internal brand name, not geography
- Label failure —
"Treatments"used where patients arrive looking for symptoms - Wayfinding gap —
"Contact"buried insideAbout Us - Information hierarchy —
"In The News"given equal nav weight to clinical sections - Missing consolidation — No Patient Resources section despite high user need
The skill proposed a full nav restructure, wrote before/after comparisons, produced an IA pattern rationale, and generated a 9-section Word report — all in one conversation. As of v2.2, the methodology behind this audit is codified in references/audit-methodology.md, so every audit follows the same evaluation lenses and reporting structure.
→ See examples/GI_Partners_IA_Audit_Report.docx
What's new in v2.2
Theme: from judgment to execution. v2.0 taught the skill what good IA looks like; v2.2 makes the audit process itself reproducible.
| New capability | Detail |
|---|---|
| Audit methodology (new reference file) | Six-stage process: Frame → Walk → Evaluate → Diagnose → Restructure → Report. Ten evaluation lenses, fixed issue-card format, severity definitions tied to framed user tasks, before/after comparison method, nine-section client report structure |
| Quantitative research analysis | Similarity matrices, dendrogram agreement levels, standardisation grids; tree-test success/directness/first-click metrics with benchmarks; participant-number guidance separating qualitative insight (5–8) from quantitative claims (30–50) |
| Healthcare IA | Generalised section: symptom → condition → treatment ladder, plain-language labelling with dual-register vocabulary, regulatory names as controlled vocabulary, structural defaults for clinic sites |
| AI-mediated seeking (contributed) | Fifth information-seeking mode — users ask AI assistants about a site instead of navigating it; IA implications for labels, vocabulary, and on/off-site consistency |
| Verification states (contributed) | Present / Absent / Unverified rule for all audit findings — never report "absent" without direct confirmation |
| Hesitation-point vocabulary capture (contributed) | The exact words users say at the moment of hesitation are the highest-value labelling vocabulary |
The three contributed items came from live audit and usability work by an external UX researcher — the skill's first community contributions. Full attribution in CHANGELOG.md.
v2.0 incorporated six additional books on top of the Spencer foundation:
| New capability | Source |
|---|---|
| Search systems — full chapter | Morville & Rosenfeld |
| Controlled vocabularies, synonym rings, thesauri | Morville & Rosenfeld |
| Faceted classification and guided navigation | Morville & Rosenfeld |
| LATCH organisational framework | Wurman via Marquis |
| Criteria matching process for categorisation | Marquis |
| Four categorical considerations (user needs / business goals / current state / future) | Marquis |
| Structural audit methodology with spreadsheet template | Marquis |
| Labelling guidelines — clarity, specificity, inclusivity, consistency | Marquis |
| Conway's Law warning | Marquis |
| Folksonomy failure modes and controlled tagging | Marquis |
| Ontology as distinct from vocabulary | Covert |
| Nouns + verbs requirements model | Covert |
| Scales of abstraction (object → ecosystem) | Covert |
| Scanning, satisficing, muddling through | Krug |
| First-click testing | Krug |
| Organising systems framework vocabulary | Glushko |
| Resource description and aboutness theory | Glushko |
| Agent-type analysis (human vs algorithmic tagging) | Glushko |
| Five cross-channel heuristics | Resmini & Rosati |
| Internal vs external consistency | Resmini & Rosati |
See CHANGELOG.md for full details.
Source literature
Built from the actual practitioner texts. Decision rules and anti-patterns extracted from source material — not summaries.
Primary source
A Practical Guide to Information Architecture — Donna Spencer (UX Mastery, 2010) Spencer's People × Content × Context model, classification scheme selector, and labelling principles form the backbone of this skill.
v2.0 additions
| Book | Author | What it added |
|---|---|---|
| Information Architecture for the Web and Beyond | Morville, Rosenfeld & Arango (4th ed.) | Search systems, controlled vocabularies, faceted classification, bottom-up IA |
| Everyday Information Architecture | Lisa Maria Marquis | LATCH, criteria matching, structural audits, labelling guidelines, Conway's Law |
| How to Make Sense of Any Mess | Abby Covert | Ontology, nouns + verbs model, scales of abstraction, "don't say" lists |
| Don't Make Me Think | Steve Krug | Scanning/satisficing/muddling, first-click testing, conventions, visual hierarchy |
| The Discipline of Organizing | Robert J. Glushko (4th ed.) | Organising systems framework, resource description, agent types, aboutness |
| Pervasive Information Architecture | Resmini & Rosati | Five cross-channel heuristics, internal/external consistency, ubiquitous ecologies |
v2.2 additions draw on designsuite.ai audit practice and community contributions from live UX research — see CHANGELOG.md.
Useful tools that pair with this skill
| Tool | Use case |
|---|---|
| Optimal Workshop | OptimalSort (card sorting) + Treejack (tree testing) — the gold standard |
| UXtweak | Card sort, tree test, usability — free tier available |
| Maze | Card sorting + first-click testing + prototype tests |
| Lyssna | Tree testing + first-click + preference tests |
| UXMetrics | Lightweight focused card sort + tree test |
| Miro | Affinity mapping, sitemap diagramming, sticky-note card sorts |
| Figma | Navigation wireframing + IA structure visualisation |
Who this is for
- Web designers and UX designers working on site structure and navigation
- WordPress / Elementor developers making navigation and menu decisions
- Healthcare and clinic websites — plain-language, symptom-first IA
- Agency teams preparing client-facing IA audit deliverables
- Content strategists building taxonomy and labelling systems
- Product designers working on search and discovery experiences
Further reading
| Book | Author | Why |
|---|---|---|
| A Practical Guide to Information Architecture | Donna Spencer | Best applied methodology — built this skill |
| Information Architecture for the Web and Beyond | Morville, Rosenfeld & Arango | Foundational vocabulary; search systems; controlled vocabularies |
| How to Make Sense of Any Mess | Abby Covert | Clearest intro to IA thinking; language-first approach |
| Everyday Information Architecture | Lisa Maria Marquis | Content-first IA; strongest on labelling, taxonomy, and audits |
| Don't Make Me Think | Steve Krug | Usability lens on navigation; billboards not books |
| The Discipline of Organizing | Robert J. Glushko (ed.) | Academic depth; bridges library science + HCI |
| Pervasive Information Architecture | Resmini & Rosati | Cross-channel IA across apps, devices, physical spaces |
| Search Patterns | Morville & Callender | Search as navigation; complements browse-first IA |
Contributing
Contributions are welcome — but keep the skill tight. The core file loads on every conversation and must stay scannable.
Write for a working designer, not an academic. Decisions over definitions. "Use X when Y" over "X is defined as…" — every addition should answer a real decision a designer faces.
What we want
- Sharper decision rules for scheme and pattern selection
- Real-world examples — IA patterns applied to healthcare, e-commerce, SaaS, intranets
- Labelling anti-patterns worth documenting
- Practical notes from using Optimal Workshop, UXtweak, Maze, etc.
- Anything factually wrong or misleading corrected
What we don't want
- Pure theory without practical application
- Long additions to
SKILL.md— it loads on every conversation - Frameworks that haven't been tested in real projects
How to contribute
1. Fork the repo
2. Edit the relevant file in skill/references/
3. Keep the tone: practitioner-first, decision-rule focused, concrete examples
4. Open a PR with a one-line description of what changed and why
v2.2 includes the skill's first merged community contributions — proof the process works.
About
Built and maintained by designsuite.ai — UI/UX studio specialising in WordPress and healthcare web design.
Our sister brand medicalsuite.ai focuses on healthcare web design globally.
We build Claude skills as internal tools and release them publicly when we think they're useful to other practitioners.
MIT License · designsuite.ai · medicalsuite.ai