What problem does it solve? Large or unfamiliar codebases accumulate architectural drift, god modules, misleading names, and hidden tech debt that no single review can surface. This Skill maps the entire system and converts findings into a durable, citable backlog instead of a one-off report. ## Core Features & Use Cases - End-to-end mapping: Builds a file manifest, dependency graph, flow traces, and a control-deck inventory of configs, env vars, and external state. - Surface integrity checks: Audits exported functions, CLI flags, API routes, and log/event strings for name-vs-behavior mismatches, near-duplicate pairs, and doc drift. - Routed backlog with stable IDs: Every finding gets a permanent ID, severity, blast radius, and a lane (plan-first, single-pr, or defer) that downstream plans and PRs can cite. - Use Case: Before a major refactor, run the audit to get audit-report.md and audit-backlog.md, then triage each item by fixing it, deferring it, or handing it to a supporting skill like python-refactor. ## Quick Start Audit this repository end to end and produce an audit report plus a routed backlog of findings with stable IDs.
