Uxcel-Lab Microcopy Audit Skill
How this skill behaves (read first)
This is an evaluative skill — it reviews existing copy, auto-runs, and is built to be called as a validation step by generative skills (the ux-microcopy writer, plus typography, dashboard, onboarding, and component skills hand off here for their labels and messages).
It owns copy quality. Deliberate deferrals:
- Manipulative copy (confirmshaming, fake urgency, guilt) → the dark-patterns audit owns intent; flag and route.
- Copy accessibility specifics (alt text rules, screen-reader behavior) → the accessibility audit; this skill flags plain-language/clarity, defers the technical a11y mechanics.
Scope discipline. When invoked directly (the user named this audit), review only this concern — don't pull in sibling audits. It runs alongside other lenses only when an orchestrator (ux-design-review) or a generative skill calls it under docs/orchestration-policy.md, where it sits in Tier A — always-relevant, auto-runs. Explicit scope always wins.
What this skill changes vs. default behavior
By default, copy review is impressionistic ("sounds fine") and partial — it misses the hidden copy (error states, empty states) and never checks consistency across a flow. This skill forces:
- A content inventory — sweep all copy, including hidden states.
- Named criteria — clarity, concision, action-orientation, consistency, tone fit, inclusivity — not vibes.
- A rewrite for each finding, plus severity.
- A consistency pass across the whole flow, which single-screen review can't do.
The audit method (content inventory)
- Inventory the copy. List every piece of text in the flow/product — and don't miss hidden content: error messages, empty states, tooltips, confirmations, notifications.
- Evaluate against the criteria below. Capture screenshots/locations so fixes are easy to apply.
- Prioritize — one step at a time; lead with what costs users most.
- Recommend rewrites, not just flags.
The criteria — what to check and what a violation looks like
1. Clarity & plain language
Flag when: jargon or high-register words; ambiguous phrasing; copy that needs re-reading.
- ❌ "Authentication credentials could not be validated."
- ✅ "That password doesn't match. Try again or reset it."
2. Concision
Flag when: filler words, multi-sentence labels, saying in 20 words what 6 would carry.
3. Action-orientation & specificity
Flag when: generic labels ("OK", "Yes", "Submit", "Start"); CTAs that don't say what happens or why.
- ❌ "Submit" → ✅ "Create account"
4. Consistency
Flag when: the same concept is named differently across the product ("scheduling" vs. "booking"), inconsistent button verbs, or platform-mismatched terms ("click" on mobile). One term per concept.
5. Tone fit
Flag when: the register is wrong for the moment — a joke on a payment error, cold/robotic copy on a welcome, hedging ("maybe") on a security action. Tone should match brand and moment.
6. Error & empty/success states
Flag when: errors state the problem but not the fix, blame the user, or use ALL-CAPS; empty states are blank instead of teaching the first action; success isn't acknowledged.
7. Inclusivity & clarity for all
Flag when: gendered pronouns where neutral works; non-descriptive links ("click here"/"More"); placeholder used as the only label. (Technical a11y → accessibility audit.)
8. Honesty
Flag when: confirmshaming, manufactured urgency, guilt, or confusing double-negatives. Note it and route to the dark-patterns audit for the intent classification.
Severity rating
| Severity | Meaning |
|---|---|
| Critical | Copy blocks or misleads — an error with no path forward, a label that causes the wrong action, manipulative copy. |
| Major | Real friction — vague CTAs, jargon, inconsistent terms for the same concept, wrong tone in a sensitive moment. |
| Minor | Polish — wordiness, a slightly off but understandable phrase. |
Audit output format
## Microcopy Audit — [Screen / Flow]
Context: [product voice · audience · the moment]
### Critical
- [Location] "current copy" → "rewrite" — why. (Criterion)
### Major
- …
### Minor
- …
### Working well ✓
- …
**Consistency notes:** terms used inconsistently across the flow.
**Top 3 rewrites:** highest-impact fixes.
End with the consistency notes and top rewrites — those are what a calling generator or writer acts on.
How to run a good audit
- Inventory first, including hidden states — that's where the worst copy hides.
- Check consistency across the whole flow, not just each screen alone.
- Rewrite, don't just flag — show the better version.
- Match tone to the moment, not a single house style.
- Defer manipulation intent to dark-patterns and technical a11y to accessibility.
Common do/don't patterns
| ❌ Don't | ✅ Do |
|---|---|
| "The copy sounds fine" | Inventory all copy (incl. hidden states); name criteria; rate severity |
| Review one screen in isolation | Check terminology consistency across the whole flow |
| Flag without fixing | Provide the rewrite |
| Pass generic "OK/Submit" labels | Flag → verb + outcome |
| Miss error/empty states | Audit the hidden copy explicitly |
| Re-derive manipulation or a11y rules | Route to dark-patterns / accessibility |