Network

mlab network

Capture every request a page makes.

mlab network https://example.com
mlab network https://example.com --json

  🌐 Network Capture  https://example.com

The URL is loaded in an instrumented browser upstream, and the command lists what that load pulled in: each request with its status, resource type, size and whatever failed.

What it is for

The third-party inventory of a page, without opening devtools and without loading the page on your own machine. What a marketing site actually calls at runtime, which CDN a script really comes from, what a checkout page reaches for before you have entered anything.

It pairs with scan url, which analyses the URL and its host statically. network answers the different question of what happens when someone opens it.

It is the slowest command in the CLI

A real browser loads the page upstream, so this takes seconds where a lookup takes milliseconds. The spinner says Loading … in a browser for exactly that reason. --timeout raises the HTTP ceiling if a heavy page needs it.

It does load the page

Server-side, not from your machine — but it is a real load, with whatever that implies for a hostile URL: it fetches the target, executes its scripts, and follows what they ask for.

That is the whole function of the command, so the point is not to be careful with it but to know what you are asking for. scan url without --resolve is the version that does not touch the target.

Output

--json gives the request list in full — the form to script against:

mlab network https://example.com --json | jq -r '.requests[].url' | sort -u
mlab network https://example.com --json \
  | jq -r '.requests[] | select(.status >= 400) | "\(.status) \(.url)"'

-o csv is refused. See Output.