Parse raw customer feedback (app reviews, support tickets, transcripts) into severity-ranked themes, scan the codebase to identify affected files, and propose 3 structural edit options per theme — minimal, refactor, or architectural
Target user: Developers and product engineers who have customer feedback and need to turn it into concrete, file-level code change proposals
cp SKILL.md ~/.claude/skills/Reload Claude Code. Trigger phrase: here's some customer feedback what should I change / I have app reviews help me prioritize what to fix / translate this user feedback into code changes
Input:
"The checkout keeps freezing on mobile" (14 mentions)
"I can't figure out how to remove items from my cart" (8 mentions)
"App crashes when I try to upload a photo" (3 mentions)
Output:
Theme 1 — Checkout freeze (UX/Bug, Critical, 14 mentions)
Option A: Add loading state guard in CheckoutButton.tsx — targeted patch
Option B: Refactor cart submission flow to async with error boundary
Option C: Extract checkout into separate route with independent state tree
Requires a codebase to scan — output is file-level proposals, not abstract advice. Works best when feedback includes concrete user actions (what they tried to do), not just outcomes (it's broken).
Built by Agent Engineer Master — production-ready Claude Code skills on demand. Every skill is engineered to the AEM production bar: trigger logic, methodology, output contract, and edge case handling. All four. Every time.