LOCAL-FIRST PERFORMANCE LAB

See where your page actually waits.

Turn a browser network capture into a clear request waterfall, transfer breakdown, and a short list of places worth investigating.

01 No account 02 No server upload 03 Explainable metrics
START AN ANALYSIS PRIVATE BY DESIGN

Drop in a HAR file.
Get the full picture.

Export a HAR from your browser’s Network panel. The file is read in this tab; it is never sent to a server.

Up to 20 MB / 10,000 requests · URLs are shown without query strings
THE OVERVIEW

Network, at a glance.

Loading demo…

Loading the synthetic example…

01 / VOLUME—RequestsCaptured entries
02 / WEIGHT—Known transferFrom recorded wire-size fields
03 / SPEED—p95 request durationNearest-rank percentile
04 / WINDOW—Capture spanFirst start to last finish
BEFORE / AFTER

Compare a second capture.

Use two captures of the same page and conditions. Files stay in this tab; only aggregate, sanitized data is exported.

YOUR THRESHOLDS

Performance budgets

Adjust these to fit your own page. A red result is a prompt to investigate, not a diagnosis.

WHAT STANDS OUT

Investigation queue

Heuristic signals from the capture, with no invented root-cause claims.

FOLLOW THE REQUESTS

The waterfall

Every bar starts when the request began; its width is the recorded duration.

Document Script Style Image Data Other
0 requestsTimeline
RequestHostStatusTransferDurationTimeline

Timings come from the HAR export. Browser phase fields may be missing or estimated; a waterfall alone cannot prove a critical path.

SELECTED REQUEST

Request detail

TRANSFER DISTRIBUTION

By host

Known transfer only. The first observed host is labeled primary, not necessarily the site's origin.

THE METHOD

Useful evidence.
Honest limits.

01Read locallyYour chosen file stays in browser memory. HarFlow makes no upload request.
02Normalize carefullyQuery strings, fragments, credentials, and request/response headers are omitted from the interface and export.
03Investigate, then verifyBudget misses and hints are leads. Confirm changes with repeat captures and lab or field metrics.