Skip to content

Features Overview

The PageSpeed Insights MCP server provides six workflow-oriented tools designed to help developers and AI assistants analyze, diagnose, compare, and optimize web performance.

📊 Performance Analysis (pagespeed_analyze_page)

Run comprehensive Lighthouse audits across configurable categories and device strategies:

  • Multi-Category Audits: Audit Performance, Accessibility, Best Practices, SEO, and PWA scores.
  • Core Web Vitals: In-depth lab metrics including LCP (Largest Contentful Paint), CLS (Cumulative Layout Shift), and INP (Interaction to Next Paint) / TBT (Total Blocking Time).
  • Multi-Run Stability: Execute 1 to 5 runs to calculate the median run and spread, reducing lab variability.
  • Tailored Report Types:
    • summary: Concise overview with scores, Core Web Vitals, and top opportunities.
    • full: Complete Lighthouse audit breakdown.
    • recommendations: Prioritized action plan with estimated byte and time savings.
    • audit: Detailed non-performance audit findings (accessibility, SEO, best practices).
    • performance-map: Visual flowchart using Mermaid syntax illustrating performance relationships.

🔍 Targeted Diagnostics (pagespeed_diagnose_page)

Instead of dumping massive JSON trees, investigate performance bottlenecks through seven focused diagnostic lenses:

  • visual: Filmstrip loading progression and final page screenshots.
  • elements: Specific DOM elements triggering layout shifts (CLS) or slow Largest Contentful Paint (LCP).
  • network: Resource transfer sizes, compression, HTTP request waterfalls, and latency bottlenecks.
  • javascript: Main-thread script execution time, boot-up evaluation overhead, and long tasks.
  • images: Unoptimized images, missing responsive formats (AVIF/WebP), sizing mismatches, and layout shifts.
  • render-blocking: Critical CSS and synchronous scripts delaying First Contentful Paint.
  • third-parties: Impact of external tags, trackers, analytics, and widgets on page weight and execution.

🌍 Real-World User Field Data (pagespeed_get_field_data)

Lab data (Lighthouse) simulates visits under synthetic network conditions. Real-user metrics come from the Chrome User Experience Report (CrUX):

  • 75th Percentile (p75) Metrics: Real-user distributions for LCP, CLS, INP, and FCP.
  • Dual Query Scopes:
    • scope=page: Targeted field data for a specific URL.
    • scope=origin: Aggregate field metrics across the entire origin (domain), essential for newly launched pages or URLs with low traffic.
  • Device Form Factors: Filter by PHONE, DESKTOP, TABLET, or ALL (origin scope).

⚖️ Comparison & Regression Baselines (pagespeed_compare_pages)

Track performance deltas across deployments and environments:

  • Side-by-Side Comparison (mode=pages): Compare two URLs (e.g. Staging vs. Production, or competitor benchmarking) across all Core Web Vitals.
  • Baseline Tracking (mode=baseline): Store an initial performance baseline locally and measure subsequent runs against it. Accurately verify whether code optimizations or dependency updates improved performance or caused regressions.

📦 Batch Analysis (pagespeed_analyze_batch)

Analyze up to 10 URLs in parallel with progress notifications, returning per-page results alongside aggregate pass/fail statistics. Ideal for auditing top landing pages or sitemaps.

🧹 Cache Management (pagespeed_clear_cache)

The server caches API responses in-memory with a configurable TTL (default: 1 hour) to preserve API quota. Use pagespeed_clear_cache to immediately invalidate the local cache after deploying changes, forcing fresh measurements from Google.