agentsclimarketplace

Copy coach

Skill nhruska/nhruska.github.io/.claude/skills/copy-coach

Microcopy coach for the Music app - button labels, modal text, empty states, tips, and error messages, level-appropriate and jargon-honest. Use BEFORE writing or changing ANY user-visible string (the S-PERSONA-COPY goalpost owns the Solo-modal rewrite). Summoned per the SME dynamic-summoning rule.From its SKILL.md

Install
npx -y skills add nhruska/nhruska.github.io --skill copy-coach

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

2.8 KB, as published. Nobody here has run it

Copy coach

Every string is UI. The app's copy failures found in UAT were vocabulary failures ("Save to add a video backing track or skip to practice" - a beginner knows none of those nouns), not grammar failures.

Rules

  • Vocabulary budget by level. Beginner copy uses only: chord, song, key, play, practice, tune. Intermediate adds: progression, scale, transpose, solo. Advanced adds: mode, degree, pentatonic, diatonic. A string shown to ALL levels uses the BEGINNER budget - or becomes level-conditional (LEVELS gate).
  • Verb-first buttons, outcome-named. "Save & open Studio" fails twice: Save is the mechanism, Studio is internal jargon. Name the OUTCOME: "Keep it + practice over a track" / "Just practice" / "Keep building". Buttons say what HAPPENS, not where you go in our architecture.
  • Empty states are starters, never apologies. "No curated video yet - pick a genre and feel below" describes our gap. A starter offers the next TAP: one concrete action button. If an empty state has no button, it is a dead end.
  • Questions only when the user must choose. Never rhetorical questions in chrome ("Ready to jam?"). The one sanctioned ask-shape is a choice card (the guidance ask) - options are answers, not links.
  • Numbers beat adverbs. "so close to fitting" energy: say "6 chords", "5 taps", "2 more" - never "a few", "almost".
  • App-internal names never leak (Studio, Notable, pack, KeyExplorer, seq) unless the surface teaches that name deliberately.
  • ASCII in code, typography in UI. UI strings may use real punctuation, but match the app's existing style (hyphens, straight quotes); accidentals use the app's glyph conventions (♭/# per surface precedent).

The rewrite loop (USDD-shaped)

  1. Identify the audience level(s) of the surface.
  2. Rewrite inside that vocabulary budget; outcome-named verbs.
  3. Encode as a persona scenario assertion (the copy IS testable text).
  4. Read it aloud as the persona: would Sam-the-beginner know every noun?

Self-check

  1. Which level reads this, and is every noun inside their budget?
  2. Does the primary button name an outcome the user wants?
  3. Does any empty state lack a next-tap button?
  4. Did an internal architecture name leak?

Related

  • pedagogy-coach - when to say anything at all
  • ux-coach - visual emphasis of what the copy labels
  • ux-friction-profiles G1 (S-PERSONA-COPY) - the first build this skill drives

Keep looking

Skills are one crate of 608,475. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.