web-vitals-inp-deep
Deep INP (Interaction to Next Paint) testing: decomposes input delay, processing duration, and presentation delay via the web-vitals/attribution build, asserts per-interaction INP budgets in Playwright using PerformanceObserver plus the web-vitals visibilitychange flush, and identifies long tasks blocking the main thread. Use when a page feels unresponsive while LCP and CLS are green, or to gate key interactions (form submit, modal open, route change) under an INP budget in CI. Covers interactions only - for page-load Web Vitals gating use lighthouse-perf; for service-worker cache-strategy latency use the qa-pwa plugin's service-worker skills.
Install with skills.sh (any agent)
npx skills add testland/qa --skill web-vitals-inp-deepweb-vitals-inp-deep
INP is "the time from the start of the interaction to the moment the next frame is fully presented" per the INP web.dev article (opens in new window). Thresholds at the 75th percentile of page loads:
When to use
Step 1 - Install the web-vitals library
npm install web-vitalsPer the INP web.dev article (opens in new window).
Step 2 - Measure INP in lab (per interaction)
import { onINP } from 'web-vitals';
onINP((metric) => {
console.log('INP:', metric.value, metric);
// metric.attribution gives interaction details (target, eventType, etc.)
});Per the INP web.dev article (opens in new window): the web-vitals library handles edge cases (BFCache restoration, visibility changes) that raw PerformanceObserver does not.
Step 3 - Decompose INP
INP encompasses three components per the INP web.dev article (opens in new window):
Use metric.attribution (provided by web-vitals INP attribution build) to identify which component dominates:
import { onINP } from 'web-vitals/attribution';
onINP((metric) => {
console.log('Total INP:', metric.value);
console.log('Input delay:', metric.attribution.inputDelay);
console.log('Processing duration:', metric.attribution.processingDuration);
console.log('Presentation delay:', metric.attribution.presentationDelay);
console.log('Long animation frame:', metric.attribution.longAnimationFrameEntries);
});Step 4 - Playwright assertion per interaction
import { test, expect } from '@playwright/test';
test('modal open INP under 200ms', async ({ page }) => {
await page.addInitScript({ path: 'node_modules/web-vitals/dist/web-vitals.iife.js' });
await page.goto('https://localhost:3000');
await page.evaluate(() => {
(window as any).__inpValues = [];
(window as any).webVitals.onINP((m: any) => {
(window as any).__inpValues.push(m.value);
});
});
// The interaction under test
await page.click('[data-testid="open-modal"]');
await page.waitForSelector('[role="dialog"]');
// Force INP to flush. web-vitals finalizes INP inside its own
// visibilitychange handler, which reads document.visibilityState
// synchronously (per the [web-vitals README]). Redefine the property to
// 'hidden' FIRST, THEN dispatch the event: if you dispatch first, the
// handler still sees 'visible' and never reports.
await page.evaluate(() => {
Object.defineProperty(document, 'visibilityState', {
configurable: true,
get: () => 'hidden',
});
document.dispatchEvent(new Event('visibilitychange'));
});
const inps = await page.evaluate(() => (window as any).__inpValues);
const max = Math.max(...inps);
expect(max).toBeLessThan(200);
});Step 5 - Identify long tasks blocking main thread
test('no long tasks > 50ms during route change', async ({ page }) => {
// Install the observer BEFORE navigation and push entries onto a
// window-scoped array. A closure-local array would never reach
// window.__longTasks, so the later read returns [] and the assertion
// passes vacuously regardless of real long tasks.
await page.addInitScript(() => {
(window as any).__longTasks = [];
const obs = new PerformanceObserver((list) => {
for (const e of list.getEntries()) {
(window as any).__longTasks.push({
name: e.name, duration: e.duration, startTime: e.startTime,
});
}
});
// 'longtask' is the Long Tasks API entry type; buffered:true replays
// tasks recorded before the observer attached (per [MDN longtask]).
obs.observe({ type: 'longtask', buffered: true });
});
await page.goto('https://localhost:3000');
await page.click('[data-testid="route-link"]');
await page.waitForLoadState('networkidle');
const blocking = (await page.evaluate(() => (window as any).__longTasks ?? []))
.filter((t: any) => t.duration > 50);
expect(blocking).toEqual([]);
});The richer successor to Long Tasks is the Long Animation Frames (LoAF) API (type: 'long-animation-frame'), but it is not yet Baseline across browsers per MDN LoAF (opens in new window), so attach it as a separate, guarded observer rather than relying on it alone.
Step 6 - CrUX field data correlation
Lab + field don't always match. Pair lab tests with CrUX queries:
# CrUX REST API
curl -X POST 'https://chromeuxreport.googleapis.com/v1/records:queryRecord?key=<API_KEY>' \
-H 'Content-Type: application/json' \
-d '{"url":"https://yoursite.com/","metrics":["interaction_to_next_paint"]}'Field 75th percentile INP ≤ 200 ms = "Good" per the INP web.dev article (opens in new window). Lab passing + field failing = sample population mismatch (real devices slower, real interactions less predictable).
Anti-patterns
| Anti-pattern | Why it fails | Fix |
|---|---|---|
Use raw PerformanceObserver for INP | Misses BFCache, visibility-change reporting | Use web-vitals library (Step 2) |
Test INP only on click events | INP measures all interactions; tap, keypress also count | Test representative interactions per type |
| Use desktop CPU 4× speed | Hides regressions; field is mobile + slower CPU | Throttle CPU 4 - 6× in Playwright config |
| Assert single sample | INP is jittery; one bad sample fails CI | Run interaction N times; assert P75 ≤ budget |
Fix INP by fragmenting handlers with setTimeout(0) | Hides yield, doesn't reduce work | Actually reduce processing duration via deferral patterns |
Limitations
References
Related skills
db-query-plan-analyzer
Reads `EXPLAIN` / `EXPLAIN ANALYZE` output from PostgreSQL, MySQL, or SQLite - identifies the dominant cost (sequential scan, nested loop, sort spill, missing index, type-cast preventing index use), proposes the specific index or query rewrite to fix it, and emits the candidate `CREATE INDEX` statement. Use when load testing or production telemetry shows the database as the bottleneck and the team needs targeted query-level remediation.
flame-graph-analyzer
Reads CPU flame-graph output from py-spy (Python), async-profiler (JVM), Go pprof, or Node.js `perf_hooks` / clinic.js: identifies the hot path (top sample-time frames), classifies the bottleneck (CPU-bound vs lock contention vs allocator pressure), and proposes the next investigation step. Use when a perf regression is bisected to a commit but the hot path inside it is unclear; for tail-latency percentiles use the latency-percentiles reference in k6-load-testing, and for a slow SQL hot path use db-query-plan-analyzer.
jmeter-load-testing
Authors Apache JMeter `.jmx` test plans (Thread Groups + HTTP samplers + assertions + listeners) in the JMeter GUI, runs them headlessly via `jmeter -n -t plan.jmx -l results.jtl`, generates an HTML dashboard with `-e -o`, and gates CI on JTL parsing. Use when the project has an existing JMeter investment, needs JVM-native load tooling, or works in domains with strong JMeter community support (banking, telecom, enterprise).
k6-load-testing
Authors k6 JavaScript load-test scripts (VU loops + checks + sleeps), configures the `options` block with `stages` (ramp-up patterns) and `thresholds` (p(95) latency, error rate), runs via `k6 run script.js` or `--vus / --duration` ad-hoc flags, and uses thresholds as the CI pass/fail signal. Includes a latency-percentile interpretation reference: tail ratio (p99/p50), bimodal-distribution detection, coordinated omission and why naive p99 is optimistic, and constant-vus vs constant-arrival-rate executors. Use when the project ships HTTP / WebSocket / gRPC load tests and the team wants developer-friendly JavaScript authoring, or when a k6 threshold passes but the system still feels slow.
lighthouse-perf
Configures Lighthouse CI (`@lhci/cli`) to audit Web Vitals (LCP, INP, CLS) on every PR, asserts against canonical thresholds (LCP ≤2.5s, INP ≤200ms, CLS ≤0.1 at the 75th percentile), uploads Lighthouse reports as build artifacts, and posts deltas as PR comments. Includes a budget-authoring reference: per-route LCP/INP/CLS thresholds by traffic class (cached / dynamic / api-heavy / form-heavy / media-heavy) via `assertMatrix`, plus `budget.json` resource-size caps (JS / CSS / images / total bytes). Use when the project ships a web frontend and the team needs continuous Web Vitals monitoring tied to PR gating, or needs its first Lighthouse budgets drafted.
load-testing-overview
Teaches load and performance testing from zero: a tool-selection table choosing between k6, JMeter, Gatling, Locust, and Artillery from observable project facts; the six load profiles (smoke, average-load, stress, spike, soak, breakpoint); open vs closed workload models; why percentiles beat averages; turning a run into a pass/fail CI gate with a first runnable k6 script; a performance-incident triage workflow (confirm with a k6 smoke run, flame-graph the hot path, check slow queries, localize the cause); and full Gatling (Simulation DSL, injectOpen/injectClosed, setUp().assertions()) and Locust (HttpUser + @task locustfile, headless / distributed runs, CSV gating) deep dives in references. Use when a service needs performance coverage and the tool, load profile, or pass/fail threshold has not been decided yet, or when a live performance incident needs cause localization.
perf-budget-gate
Builds a unified release-readiness gate that aggregates verdicts from any combination of k6 / JMeter / Gatling / Locust load runners and Lighthouse CI Web Vitals, applies severity-aware pass/fail thresholds, and emits a single go / no-go decision with per-metric deltas vs the main-branch baseline. Posts the delta as a PR comment when the team has the integration set up. Use when authoring a CI step that gates a deployment on cross-runner perf compatibility.
slo-load-test-plan
Turns a service's SLOs and endpoint traffic mix into a named scenario matrix: one scenario per SLO boundary condition, a load profile (smoke, average-load, stress, soak, spike, breakpoint) per scenario, an open or closed workload injection model, a threshold expression derived from the SLO the scenario guards, and an error-budget calculation that sets the soak run's failure allowance. Stays runner-agnostic and fixes the pass/fail line before any tool is configured. Use when an SLO document and an endpoint list both exist but nobody has decided which load runs to make, what shape of load each carries, or what number would count as a failure.