Help us improve
Share bugs, ideas, or general feedback.
From cms-cultivator
Runs a strategist-focused UX site audit using UX laws, content hierarchy review, qualitative data synthesis, and Lighthouse. Produces Markdown summary and HTML artifact for client sharing.
npx claudepluginhub kanopi/claude-toolbox --plugin cms-cultivatorHow this skill is triggered — by the user, by Claude, or both
Slash command
/cms-cultivator:strategist-site-auditThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
A strategist-, UX-lead-, and PM-facing audit skill for discovery and pre-discovery phases. Produces presentation-ready, client-safe deliverables — not developer technical reports.
Audits deployed websites and web apps post-launch, cycling through Fix-Audit-Test phases to identify and resolve issues.
Reviews websites exhaustively for AEO, GEO, SEO, UI/UX psychology, and copywriting. Crawls multiple pages, takes screenshots, scores elements, and provides charts with actionable fixes.
Audits landing pages for conversion optimization, scoring above-fold clarity, trust signals, form friction, message match, page speed, and mobile UX with prioritized recommendations.
Share bugs, ideas, or general feedback.
A strategist-, UX-lead-, and PM-facing audit skill for discovery and pre-discovery phases. Produces presentation-ready, client-safe deliverables — not developer technical reports.
This skill drives a browser via CoWork and writes two artifacts to disk (a Markdown summary and an HTML report). It does not push code, post anywhere, or modify the audited site. The HTML Artifact is generated as a draft — the strategist iterates on tone and emphasis before sharing.
Not for developers. Technical audits (WCAG conformance, OWASP, Core Web Vitals deep-dive) belong to accessibility-audit, security-audit, performance-audit, or live-site-audit. This skill's output is client-readable, jargon-light, and frames findings as opportunities — not bug tickets.
Without CoWork, the skill cannot capture screenshots or run Lighthouse automatically. Fallback: the strategist provides screenshots manually and runs Lighthouse via PageSpeed Insights. Document this limitation up front if CoWork isn't connected.
The skill accepts (and will explicitly ask for) any of the following qualitative data the strategist already has:
If none are provided, the skill runs from site observations alone and notes the absence of qualitative context as a discovery follow-up.
Before navigating anything, ask the strategist:
For each page in scope, capture:
Catalogue each screenshot with a clear filename and a short caption that will appear in the HTML Artifact and the Markdown Summary. Save to a working directory (e.g., ./strategist-audit-screenshots/<domain>/).
Apply all 21 core laws from Jon Yablonski's Laws of UX:
| # | Law | What to look for |
|---|---|---|
| 1 | Jakob's Law | Conventions match user expectations from similar sites |
| 2 | Fitts's Law | Target size and distance — primary CTAs large enough and near user attention |
| 3 | Hick's Law | Choice overload in nav, filters, forms; decision time vs. option count |
| 4 | Miller's Law | Chunked information — no list with > 7 items without grouping |
| 5 | Peak-End Rule | Strongest impressions: hero moment and final CTA / thank-you state |
| 6 | Von Restorff Effect | Distinct items stand out (one primary CTA per view, not five) |
| 7 | Aesthetic-Usability Effect | Visual polish aligned with brand and trust signals |
| 8 | Doherty Threshold | Perceived response < 400ms — loading states, skeleton screens, progressive disclosure |
| 9 | Law of Proximity | Related elements grouped; unrelated elements separated |
| 10 | Law of Similarity | Visually similar elements perceived as related |
| 11 | Law of Common Region | Bordered/background-separated regions group their contents |
| 12 | Law of Uniform Connectedness | Connected elements feel related (lines, shared backgrounds) |
| 13 | Law of Prägnanz | Visual simplicity — viewers default to the simplest interpretation |
| 14 | Serial Position Effect | First and last items in lists carry more weight (nav order, content priority) |
| 15 | Zeigarnik Effect | Incomplete tasks linger in attention — progress indicators, save-for-later states |
| 16 | Tesler's Law | Conservation of complexity — system absorbs complexity so users don't have to |
| 17 | Postel's Law | Be liberal in input acceptance (forms forgiving of formatting variations) |
| 18 | Goal-Gradient Effect | Progress acceleration — visible progress toward a goal increases completion |
| 19 | Occam's Razor | Simplest design solution that satisfies the requirement |
| 20 | Pareto Principle | 80% of effects from 20% of causes — focus on the high-impact issues |
| 21 | Parkinson's Law | Work expands to fill available time — constrain form fields, decision points |
For each violation found, record:
Do not pad findings. If a law is observed correctly (positive pattern), note it under "What's working" — those are as useful for strategy decks as violations.
For each page in scope, evaluate:
Capture findings per page, with specific recommendations (e.g., "Move pricing above the testimonials block — pricing is currently below the fold on mobile and is the user's primary question on this page").
If the strategist provided qualitative inputs (Step 1.3), weave them through the findings:
If no qualitative data was provided, note the absence as a discovery activity to schedule (interviews, heatmap install, survey design).
For each page in scope, capture:
accessibility-audit next)Present scores as gauges in the HTML Artifact. Frame issues in client-friendly language (e.g., "Images are slow to load on mobile" instead of "LCP > 2.5s").
Synthesise across pages:
Save to ./strategist-audit-<domain>-knowledge-summary.md. Structured for the strategist to paste into their Claude Desktop Project for the client.
# Strategist Site Audit — <Client Name> (<domain>)
**Date:** <YYYY-MM-DD>
**Pages audited:** <list>
**Qualitative data sources:** <list, or "None provided — see Discovery Follow-ups">
---
## Executive Summary
[3–5 sentences: the most important takeaways. Tone: confident, specific, non-prescriptive.
"The site shows strong consistency in visual brand application; the largest opportunities
sit in [primary CTA hierarchy], [mobile navigation depth], and [content prioritisation on
the Services page]."]
---
## What's Working
- [Positive pattern with specific example]
- [Positive pattern with specific example]
- [Positive pattern with specific example]
---
## Highest-Impact Opportunities (Prioritised)
### 1. [Opportunity title]
**Severity:** High
**Pages affected:** [list]
**UX laws involved:** [list]
**Recommendation:** [specific, actionable]
**Why it matters:** [user impact]
### 2. [Opportunity title]
[...]
---
## UX Laws Audit — Findings by Law
### Jakob's Law
- **Status:** [Met / Partially Met / Violated]
- **Observations:** [...]
- **Recommendations:** [...]
- **Screenshot:** ./strategist-audit-screenshots/<domain>/<filename>.png
[... repeat for all 21 laws ...]
---
## Content Hierarchy Review — by Page
### Homepage
- **Visual hierarchy:** [observations]
- **Content priorities:** [observations]
- **Reading patterns:** [observations]
- **Messaging clarity:** [observations]
- **Recommendations:** [specific list]
[... repeat for each page in scope ...]
---
## Lighthouse Snapshot
| Page | Performance | Accessibility | Best Practices | SEO | Top Issues |
|------|-------------|---------------|----------------|-----|------------|
| Home | 78 | 84 | 92 | 91 | Slow image load on mobile; missing alt text on hero |
| ... | | | | | |
---
## Qualitative Data Synthesis
[If qualitative data was provided: how it aligned with or contradicted observations. If not: list of recommended discovery activities to fill the gaps.]
---
## Discovery Follow-ups
- [Activity 1 — e.g., "Schedule 6 user interviews focused on Services page flow"]
- [Activity 2 — e.g., "Install heatmap tracking on top 5 pages for 2 weeks"]
- [Activity 3 — e.g., "Run tree test on proposed navigation"]
---
## How to Use This Document
Paste this entire document into the client's Claude Desktop Project as a knowledge file. Future questions like:
- "What were the Fitts's Law violations we found on Services?"
- "Summarise content hierarchy issues before I write the discovery deck"
- "What did the survey data confirm vs. contradict?"
— will be answerable in any session for this client.
Save to ./strategist-audit-<domain>.html. A polished, self-contained, print-safe report designed for client sharing.
Required sections in the rendered HTML:
Styling requirements:
After rendering, tell the strategist:
"Here's the first draft of the HTML report. You can iterate before sharing — for example:
- 'Make the recommendations more concise'
- 'Soften the tone in the executive summary'
- 'Add the client's brand color #XXXXXX as the accent'
- 'Remove the appendix — keep the report focused on findings'
The Project Knowledge Summary is also ready:
./strategist-audit-<domain>-knowledge-summary.md. Paste it into the client's Claude Desktop Project."
After delivery, suggest:
accessibility-audit if the Lighthouse accessibility score flagged real issueslive-site-audit for the technical companion audit (performance + security + code quality) — different audience, different deliverablefrd-generator to translate audit findings into a discovery FRD if scope and budget have been confirmedIf CoWork is not connected:
accessibility-audit for WCAG 2.1 conformance reportingsecurity-audit for OWASP scanningperformance-audit for Core Web Vitals optimisation with specific code fixeslive-site-audit for the developer-facing parallel auditThe strategist-site-audit complements those: it sits earlier in the engagement (discovery), serves a different audience (strategy/client), and produces a different deliverable (presentation-ready, not implementation-ready).