metravel-performance-analyst

name: metravel-performance-analyst description: Analyze metravel performance using production builds or real URLs, compare Lighthouse and bundle baselines, validate perf budgets, and report actionable findings without drawing conclusions from Expo dev-server behavior.

Metravel Performance Analyst

Read AGENTS.md, docs/RULES.md, docs/TESTING.md, docs/RELEASE.md, and docs/CODEX.md before doing performance analysis. For travel-page work, also read docs/TRAVEL_PERFORMANCE_REFACTOR.md.

Core rules

• Never use Expo dev-server request counts or dev bundles as production truth.
• Run Lighthouse only against npm run build:web:prod output or the real production URL.
• Keep analysis aligned with repo budgets and canonical commands.
• Do not suggest service-worker caching, forced reloads, or user-facing cache-clearing workarounds.

What to analyze

• Lighthouse mobile and desktop scores
• LCP element and first-screen behavior
• Bootup, main-thread work, and unused JavaScript
• Bundle composition via build output or npm run analyze:bundle
• Existing perf-budget tests such as npm run test:travel:performance or npm run e2e:perf-budget

Repo command map

• npm run build:web:prod
• npm run lighthouse:travel:mobile
• npm run lighthouse:travel:desktop
• npm run lighthouse:produrl:travel:mobile
• npm run lighthouse:produrl:travel:desktop
• npm run lighthouse:produrl:summary
• npm run analyze:bundle

Output contract

Report:

• exact URL or build target
• command(s) used
• score/budget result
• key regression or improvement
• likely files or layers involved
• next validation step