optimize

What problem does it solve? Open-ended questions like "is there room for optimization" never terminate because a model can always invent new work. This Skill splits optimization audits into two bounded branches—a closed cleanup checklist and an externally-anchored benchmarking review—so audits have objective stopping criteria. ## Core Features & Use Cases - Closeout (subtractive) audit: Scans a fixed 7-item checklist (dead code, multiple sources of truth, split implementations, defensive patches, polling, magic values, vague naming) over the current Git diff scope, producing P0/P1/P2 findings with evidence, impact, fix, and verification steps. - Elevate (additive) audit: Researches official docs, competitor source code, and industry reports via web search to find cited gaps between your implementation and current best practices, ranked by benefit and cost. - Use Case: Before merging a feature branch, ask whether the diff is clean; the Skill reports scoped findings with file and line evidence, or explicitly concludes "audit passed, no closeout points" instead of inventing work. ## Quick Start Ask the agent to audit the current Git diff for cleanup issues before merging, or to compare a module against industry best practices.