BrowserSift

Understand what broke across browsers

BrowserSift runs the same fixed local fixture through Chromium, Firefox and WebKit, then compares load milestones, resources, console diagnostics and findings in one report. This page shows one real recorded run from 13 Oct 2026 — no new scan starts when you open it.

Preflight validates a public URL and its DNS resolution and address safety only — no page is fetched. The scan itself always runs against the fixed local fixture.

Scope: the built-in local fixture page only. Public external URL scanning is not available in this prototype.

Checking demo availability…
Built-in example page (fixed fixture)
Loading test location…
Optional: test signed in (fixture account)
Check a URL (validation only)

DNS and address-safety checks only. It does not visit or test the website, and is not required for the demo.

Run fixtureRecorded demo

Product preview

What one real recorded run looks like

ms Chromium Firefox WebKit 0 30 60 90 120 150 Chromium Firefox WebKit
Each marker is an actual measured milestone at its absolute time from navigation start — this is NOT a stacked duration chart. Coincident milestones share a position; hover for the exact label. — recorded local fixture, 13 Oct 2026. One real recorded run per engine; no fabricated history curve.
script (585.1 KiB) link (514 B)
Known transfer sizes only. 0 resource(s) had unknown size and are excluded — this ring is not full page weight. — Chromium (156.0.8078.4), recorded bytes by resource type.
JavaScript error thrown on page interaction Horizontal page overflow at 1440px viewport Vue runtime warning/error Wasm load/compile failure Chromium WebKit
Teal = issue observed on that engine; grey = no observation on that engine (not a pass); blank = not collected. Severity is shown in the findings list. — findings across Chromium, Firefox and WebKit (WebKit is the engine; it is not Safari).

The full interactive demo with all eight sections lives atthe recorded demo workbench.

Features

Six things BrowserSift actually does

Loading timeline

Real navigation timing per engine: TTFB, DOM content loaded, load, first paint and LCP candidate within the bounded observation window.

Requests

Every recorded request with type, transfer size and duration, so resource weight and slow requests are visible instead of guessed.

Console diagnostics

Console messages, warnings, page errors and policy aborts captured per engine with actual timestamps and generated locations.

Engine comparison

Findings and metrics side by side across Chromium, Firefox and WebKit. Shared issues are shown as shared, not blamed on one engine.

Analysis context

Technology observations with evidence-only signals, declared adapter strategies and explicit "unknown" where nothing was observed.

Evidence

Collected artifacts are listed with kind and size. Protected content only loads after an explicit action — nothing is faked.

How it works

Three steps, real artifacts at each one

1. Record

The runner drives the fixture in each engine and captures navigation timing, requests, console output and capability signals in a structured report.

See the technical export

2. Compare

The report lines up the same metrics per engine, marks which findings appear in which browser, and keeps unobserved values explicitly "Not collected".

Compare in the demo workbench

3. Export

Two offline HTML editions — summary and technical — are generated from the same renderer used inside the app, with native SVG charts and no external scripts.

Summary ·Technical

Coverage

Engines covered today

EngineFamilyRecorded hostStatus in this run
ChromiumBlinkWindows (win32-x64)Completed — 156.0.8078.4
FirefoxGeckoWindows (win32-x64)Completed — 157.0
WebKitWebKitWindows (win32-x64)Completed — 27.2 (engine build, not Safari)

FAQ

Questions about scope

Does opening the demo start a new scan?

No. The demo replays one recorded run from 13 Oct 2026. Nothing is fetched or re-measured when the page opens.

What does preflight actually check?

Preflight validates a public URL and resolves its DNS. It never fetches or scans a page — only execution does that, and execution runs against the fixed local fixture, not your site.

Is WebKit the same as Safari?

No. Reports use the WebKit engine build (playwright's webkit), which is not Safari and is labelled as WebKit everywhere.

Are the performance numbers real-user metrics?

No. They come from a single initial-load sample on a fast local fixture inside an authenticated session. They are not RUM, CrUX or Lighthouse field data, and those modules are not claimed here.

How much will scans cost?

Pricing is planned at $2.99 (Quick) and $6.99 (Deep) but checkout is not active. See pricing for what each tier is intended to include.