What problem does it solve? Slow load times, high latency, and resource waste often get "fixed" by guesswork, leading to wasted effort on the wrong layer. This Skill enforces a measure-first discipline so every optimization targets the actual bottleneck with verified before/after metrics. ## Core Features & Use Cases - Structured Optimization Process: A six-step workflow covering baseline measurement, bottleneck profiling, impact estimation, targeted fixes, verification, and documentation. - Performance Budgets: Concrete targets such as load under 3s on 3G, API p95 under 500ms, initial bundle under 500KB, and mobile memory under 100MB. - Anti-Pattern Detection: Flags common mistakes like optimizing without profiling, caching without eviction strategy, and micro-optimizing off the critical path. - Use Case: When a user reports that an API endpoint feels slow, the Skill guides profiling to find the real cause (e.g., N+1 queries), applies the smallest effective fix, and re-measures p95 latency to confirm the improvement. ## Quick Start Ask the agent to profile the slow endpoint or page, identify the real bottleneck, and propose one measured optimization with before and after metrics.
