HAR Analyzer — Performance Waterfall & Audit
Drop a HAR file from any browser. Get a real waterfall (DNS / TCP / TLS / wait / receive segments), mime-type breakdown, largest-payload + slowest-request lists, and a per-regression audit (heavy JS, blocking CSS, ad sprawl, failed requests).
What it is
A HAR-file analyzer that renders a real waterfall — the kind Chrome devtools shows — and overlays the audit findings on top so you know what to fix before scrolling through 47 rows.
What it does
- Real waterfall — every request rendered with five timing segments (blocked, DNS, TCP connect, TLS handshake, wait/TTFB, receive). Bars positioned to scale against total page time.
- Stats strip — request count, total weight, total time, distinct hosts, failed requests, tracker count. Each cell color-codes (
badif over threshold). - Audit findings — pattern-matches the HAR for common regressions: main doc > 1s, page weight > 3MB, JS bundle > 500KB, > 3 tracking requests, multiple render-blocking CSS, > 6 distinct hosts, any 4xx / 5xx responses.
- Top breakdowns — largest payloads with bar chart, slowest requests with bar chart, mime-type breakdown grouped (html, css, js, image, video, font, json, xml, other).
Four fixtures pre-loaded: a fast 2-request site, a typical media site (11 reqs with trackers + fonts), a slow ad-heavy site (15 reqs, 8.5MB, 5 tracker pixels with full DNS+TLS handshakes), and a single API call.
Why this matters
Lighthouse runs are slow, opinionated, and require a URL. The HAR-based workflow works on staging behind auth, on internal pages, on captured slow-network sessions. Drop the HAR, see the picture in 200ms.
How it ships
Single HTML file, ~30KB. Zero dependencies. The waterfall renderer, timing-segment math, mime classifier, and finding generator are 400 lines of vanilla JavaScript. The fixtures themselves total 12KB of synthetic but realistic HAR data.