speed-it
by ravidsrk
Bring declared user journeys within pre-declared performance budgets, proven by measurement. Controlled baseline (measure to a metric contract, not once) → profile the real bottleneck → fix PR-per-hotspot with a mandatory before/after → re-benchmark to the metric's statistical contract → add CI regression guards, looping until every journey is within budget or parked. Use when "the app is slow", "perf sweep", "Core Web Vitals", "get under budget", or an unattended perf-hardening run. Not for a per-diff perf opinion (review-it) or post-deploy watch alone (observe via ship-it).