What problem does it solve? Codebases accumulate AI-generated slop, dead code, duplication, and complexity over time, and teams lack a structured way to measure these issues and decide which fixes are worth doing. This Skill runs deterministic audit tools, categorizes findings by severity, and estimates the effort and model tier needed for each fix before any code is touched. ## Core Features & Use Cases - Deterministic Scanning: Runs aislop on all projects and fallow on JS/TS projects to detect security issues, AI slop, complexity, dead code, and duplication. - Prioritized Findings Report: Groups issues into High/Medium/Low tiers with file and line references, fix approaches, and complexity estimates. - Model Recommendation: Maps each fix tier to an appropriate LLM (from lightweight models for trivial deletions to frontier models for architectural refactors) with escalation triggers. - Permission-Gated Execution: Never modifies code without an approved plan, fixes one category at a time, and re-scans after each pass to report score deltas. - Use Case: After a sprint of AI-assisted coding, ask for a code health audit to get a scored report of security findings, duplicated logic, and unused exports, plus a recommended model and budget tier for the cleanup work. ## Quick Start Ask the agent to audit this repository's code health and present a prioritized cleanup plan with complexity estimates before making any changes.
