Drop in a URL and we'll fetch the page + its same-origin assets and reveal what each
request is doing — DNS, connect, TLS handshake, wait-for-first-byte, download time,
server software, compression, protocol, IP, and which networks (PWHOIS) host each piece.
User-Agent
▾
Accept-Language
▾
Accept-Encodingbr, gzip, deflate
Starting…
Waking up the curl daemon…
0.0s
Up to 48 agents deployed in parallel
via curl_multi — the phase line above shows live deployment + return counts.
Stacked horizontal bar per asset. Bar length = real per-fetch wall-clock
(NOT summed across the parallel run). Look for outliers — usually one
long Wait (slow server) or Download (large payload).
By Asset Type — bytes
By Host — bytes
Compression / Encoding
Server Software
HTTP Protocol
An asset labelled just HTTP
(no version) means curl never saw a status line — the transfer
failed before any HTTP exchange completed (DNS failure, connection
reset, timeout, TLS error). The version that would have
been negotiated is unknown.
DNS Time per Host (first lookup)
Measured via the local resolver
(systemd-resolved on this host). When the same hosts
were resolved in a recent prior run, the resolver returns the
answer from its in-memory cache in well under 1 ms — that's
why values can read as 0.20 ms, 0.05 ms
etc. and not the real authoritative-lookup time. We can't flush
systemd-resolved without root privileges, and querying
external recursive resolvers directly over UDP/53 is firewalled on
this server, so this caching is unavoidable; numbers are most
meaningful on the first analysis of a previously-unseen host.
Per-Asset Detail
#
Type
URL
Host
IP / Network
Proto
Enc
Server
Status
Size
DNS
Conn
TLS
TTFB
Total
Speed
Reference Tree — which document linked each asset
Edges connect each parent document (HTML or CSS) to the assets it referenced.
Hover over a node to see its full URL. Scroll vertically for deep trees.