Real-World Speed Probe
PSI cosplay vs real users. Probe a URL, get results.
Metrics
Headers
—
Results
Stop grading vibes. We probe a URL and call BS where it counts: TTFB, cache headers, cookie bombs, DOM/script bloat, a rough LCP hint, and an edge-vs-origin sanity check. If it cheats the meter, we score it.
What it actually checks
TTFB & total time (HTML) — how long until the server coughs up bytes.
Cache status — public/max-age/s-maxage vs no-cache; also flags Set-Cookie on HTML (edge killer).
Headers — raw dump for receipts (CF/X-Cache, etc.).
HTML size — rough payload in B/KB/MB.
DOM weight — approximate tag count (a landfill warning).
Script count — external vs inline, because “just one more SDK” is how web rot happens.
LCP heuristic — looks at the first <img>: lazy-loaded without priority? We ding it. Preload? We note it.
Edge vs Origin — compare standard request vs cache-busted; big delta = edge doing work.
Asset timing (opt-in) — first CSS and first IMG TTFB for “is the critical path even reachable?”
Fake-Fast Detector Patch — Quick/Deep Mode (v2.2.0)
Headline: Speed when you want it. Depth when you need it.
What changed
- New mode switch: Quick = one lean probe for fast answers. Deep = unlocks Edge vs Origin and Asset timing so you can audit cache behavior and first-byte timing without guessing.
- UX polish: Enter-to-probe, sane dropdown contrast in the hero, clearer status text, and copy/export still one click.
Why it matters
Quick keeps you moving. Deep turns on the X-ray when it’s worth it. No ceremony, no iframes, no performance theater.
Compatibility
Plays nice with JS defer and normal CDN caching. (We keep admin-ajax uncached as always.)
How to use
Paste a URL → choose Quick or Deep → Probe → copy the summary or export JSON. That’s it.
TL;DR
- Quick = one request, faster feedback.
- Deep = Edge vs Origin + Asset timing on demand.
- Cleaner UI, same no-nonsense results.