What problem does it solve? Large codebases drift away from their documented architecture over time, accumulating duplicated logic, inconsistent error handling, and mismatched frontend/backend contracts. This Skill performs a structured, multi-axis audit of the entire repository and converts confirmed findings into tracked GitHub issues, while requiring human approval before any tickets are created. ## Core Features & Use Cases - Multi-Axis Codebase Scan: Evaluates domain/architecture alignment against CONTEXT.md and ADRs, backend deduplication (Flask routes, Supabase queries), frontend health (Expo/React Native navigation, render performance), and agent tool/widget payload contract integrity. - Human-in-the-Loop Review: Presents a prioritized health scorecard (P0/P1/P2 findings with file links and assumptions) and pauses for user feedback before creating any issues. - Wayfinder Issue Integration: Converts approved findings into GitHub issues with labels like ready-for-agent and wayfinder:task, wiring native blocked-by dependencies via the GitHub CLI. - Use Case: Before a major refactor of a fitness app's agent engine, run the audit to detect duplicated Garmin/Strava normalization logic and mismatched widget payloads, then turn the approved findings into an executable ticket roadmap. ## Quick Start Ask the AI to run a deep audit of the repository and present prioritized findings for review before creating any GitHub issues.
