What problem does it solve? AI-generated code often accumulates slop: obvious comments, over-defensive checks, dead code, needless abstractions, duplication, and oversized modules. Manually auditing a branch diff for these patterns is slow and error-prone, and careless cleanup risks silently changing behavior. ## Core Features & Use Cases - Behavior-locked cleanup: Writes regression tests first to pin current behavior before any line is removed, then verifies with quality gates (tests, lint, typecheck, security scan). - Ten categorized slop patterns: Covers stylistic, structural, hidden-cost, test-coverage, and sizing issues, including a mandatory modular refactor for files over 250 pure LOC. - Parallel batch processing: Dispatches deep worker agents in batches of 5 files, with retry and escalation handling for failed files. - Use Case: After an AI assistant generates a large feature branch, run this skill to strip slop from the diff while green tests guarantee no behavior changed, producing a structured removal report. ## Quick Start Ask the agent to remove AI slop from the current branch changes and verify the result with the project's test suite.
