vital-scans open quality ledger

nih.gov: Lighthouse — week 2026-W40

2 pages sampled by Google Lighthouse (its own headless Chrome). Scores are 0–100 (higher is better); metrics are Core Web Vitals. Download: CSV (per-page scores) · JSON (AI-ready, includes recommendations) · Priority pages CSV · Priority pages JSON.

Trend

Lighthouse category scores over 16 weeks
Performance Accessibility Best practices SEO W25W32W401000
Lighthouse category scores by week
WeekPerformanceAccessibilityBest practicesSEO
2026-W25n/an/an/an/a
2026-W26n/an/an/an/a
2026-W27n/an/an/an/a
2026-W28n/an/an/an/a
2026-W29619596100
2026-W30449596100
2026-W31439596100
2026-W32469596100
2026-W33379577100
2026-W3434937796
2026-W35399577100
2026-W36309677100
2026-W37n/an/an/an/a
2026-W38359577100
2026-W39309577100
2026-W4035967796

History & Trends → Severity, Lighthouse and other measures across 16 weeks.

Performance impact

How far pages fall short of Google's "good" benchmarks: Largest Contentful Paint ≤ 2.5s and page weight ≤ 1.6 MB. Lower is better.

Avg extra LCP over 2.5s
5.4s 2/2 pages over
Avg extra weight over 1.6 MB
0.00 MB 8/233 pages over

Set page_loads_per_week for this target to also estimate total wasted time and data (the way daily-dap uses traffic counts).

Medians across sampled pages

Performance
35/100
Accessibility
96/100
Best practices
77/100
SEO
96/100
Agentic browsing
78/100
Largest Contentful Paint
7.9s
First Contentful Paint
3.3s
Speed Index
4.8s
Total Blocking Time
1.3s
Cumulative Layout Shift
0

Recommendations

Issues Lighthouse flagged across the 2 sampled page(s), beyond the headline scores — grouped by category and ranked by how many pages they affect. Estimated savings (transfer bytes or load time) are Lighthouse's own estimates where available. Accessibility audits are intentionally omitted; they overlap with the axe-core findings on the Accessibility page.

Performance

Performance recommendations from Lighthouse, by sampled pages affected.
RecommendationPagesEst. saving
Reduce JavaScript execution time 2/2 2.8s
Minimize main-thread work 2/2 2.7s
Reduce unused JavaScript 2/2 697 KB · 1.9s
Reduce unused CSS 2/2 222 KB · 1.7s
Avoid multiple page redirects 2/2 1.8s
Document request latency 2/2 0.4s
Minify JavaScript 2/2 37 KB · 0.1s
First Contentful Paint 2/2 —
Largest Contentful Paint 2/2 —
Speed Index 2/2 —
Total Blocking Time 2/2 —
Max Potential First Input Delay 2/2 —
Time to Interactive 2/2 —
Forced reflow 2/2 —
Network dependency tree 2/2 —
Render-blocking requests 1/2 0.7s
Initial server response time was short 1/2 0.3s
Use efficient cache lifetimes 1/2 0.1s
Legacy JavaScript 1/2 0.1s
Font display 1/2 0.1s
Improve image delivery 1/2 0.1s
Minify CSS 1/2 2 KB
Layout shift culprits 1/2 —

Best Practices

Best Practices recommendations from Lighthouse, by sampled pages affected.
RecommendationPagesEst. saving
Browser errors were logged to the console 2/2 —
Uses deprecated APIs 2/2 —

SEO

SEO recommendations from Lighthouse, by sampled pages affected.
RecommendationPagesEst. saving
Document does not have a meta description 1/2 —

Agentic (AI-readiness)

Agentic (AI-readiness) recommendations from Lighthouse, by sampled pages affected.
RecommendationPagesEst. saving
WebMCP tools registered 2/2 —
WebMCP form coverage 2/2 —
WebMCP schemas are valid 2/2 —
llms.txt follows recommendations 2/2 —
Cumulative Layout Shift 1/2 —

What the Agentic score means

The Agentic (AI-readiness) score is new in Google Lighthouse (13.4+). It measures how well a page works for AI agents and assistants — the tools that increasingly mediate how people find and use government services. It checks for things like an llms.txt file (a machine-readable guide for language models), valid structured data, a well-formed accessibility tree that agents can parse, and WebMCP tool/form descriptions. A higher score means an AI assistant is more likely to understand the page and complete tasks on a citizen's behalf correctly.

It is experimental and evolving; treat it as a forward-looking signal, not a compliance requirement. Median across sampled pages: 78/100.

Gaps found on the sampled pages: WebMCP tools registered (2), WebMCP form coverage (2), WebMCP schemas are valid (2), llms.txt follows recommendations (2), Cumulative Layout Shift (1).

Per-page results

Lighthouse scores and Core Web Vitals per sampled page (2026-W40). Agentic = experimental agentic-browsing score. PWA / offline readiness is tracked in the Standards tab (Lighthouse 12+ removed the PWA category). — click a column heading to sort
/news-events/nih-research-matters/how-age-alters-sepsis-outcomes3995771001003.7s11.3s5.9s1.1s0
/brain/funding/dissemination-supplements30967792562.8s4.5s3.7s1.5s0.58