---
name: mcp-web-performance-auditor
description: >
  Audit and optimize web UI loading speed, responsiveness, visual stability, delivery, and perceived performance across modern and legacy stacks. Use for Core Web Vitals, Lighthouse or RUM interpretation, performance budgets, image/font/CSS delivery, caching, third-party scripts, SSR/SPA navigation, bfcache, static repository inspection, and conservative performance migration. Use mcp-ui-motion-designer for a localized animation defect, mcp-react-performance-reviewer for React/Next code-pattern review, and mcp-frontend-performance-debugger when a reproducible runtime symptom requires traces, lifecycle ownership, leak analysis, or soak testing.
---

# MCP Web Performance Auditor

Ottimizzare la user experience misurando prima, attribuendo il costo e applicando il cambiamento minimo efficace.

## Workflow

1. Definire route, journey, device, rete, browser, build e obiettivo misurabile.
2. Eseguire `scripts/mcp-web-performance-auditor/inspect_web_performance.py <project>` per inventario statico.
3. Separare lab, field/RUM e percezione utente; non confrontare misure non equivalenti.
4. Attribuire il costo a rete, immagini/font/CSS, JavaScript, rendering, terze parti o caching.
5. Definire budget e priorita prima di modificare codice.
6. Applicare patch conservative e coerenti con lo stack.
7. Ripetere la stessa matrice di misura e verificare regressioni funzionali e accessibili.
8. Demandare trace, leak e soak sistemici a `mcp-frontend-performance-debugger`.

## Regole permanenti

- Non dichiarare un miglioramento da un solo run.
- Non confondere punteggio sintetico con esperienza reale.
- Non lazy-loadare contenuto above-the-fold critico senza evidenza.
- Non rimuovere funzionalita o accessibilita per migliorare una metrica.
- Definire ownership, TTL e invalidazione per ogni cache.
- Mantenere fallback e resilienza per script e risorse di terze parti.

## Output

1. scenario e baseline
2. evidence matrix lab/field/static
3. bottleneck attribution
4. prioritized minimal changes
5. before/after validation
6. accessibility and resilience checks
7. residual runtime investigations

## References

- [accessibility-and-perceived-performance](references/accessibility-and-perceived-performance.md)
- [core-web-vitals](references/core-web-vitals.md)
- [frameworks-rendering-and-navigation](references/frameworks-rendering-and-navigation.md)
- [images-fonts-css-and-media](references/images-fonts-css-and-media.md)
- [javascript-main-thread-and-inp](references/javascript-main-thread-and-inp.md)
- [legacy-migration](references/legacy-migration.md)
- [loading-network-and-cache](references/loading-network-and-cache.md)
- [measurement-and-baselines](references/measurement-and-baselines.md)
- [official-sources](references/official-sources.md)
- [testing-observability-and-budgets](references/testing-observability-and-budgets.md)
- [third-party-and-resilience](references/third-party-and-resilience.md)