Clone their website.
Not just the frontend.

A video plays one real journey across public craigslist pages — screen recordings captured signed out by a browser agent; nothing was published or submitted, and craigslist is not affiliated with and has not endorsed this page. It shows: arriving at the SF bay home page, clicking post an ad, choosing an area in the posting wizard, choosing housing offered as the posting type, choosing apts / housing for rent, and typing a posting form with fictional data — the typed form is the last action, and it changes the page without changing the address bar. Beside the browser, a tree diagram grows downward one level per step — each of five page states fans out to the action taken and the actions it did not take, joined by straight arrows. The home page's two untaken depth-one nodes are labelled with plain words — free stuff, and about — the exact states other agents resume from. After the walked journey ends, the browser pane splits into two smaller browsers that walk those two pages at the same time — one searches the free section, switches its results to list view in place, and opens a listing; one opens the FAQ and lands on an answer page — and each of their clicks grows the matching branch under its labelled node, at the same size: the diagram never zooms. Search parallelises; that is what the ending shows.

This field makes no request. The URL only travels if you submit a clone request.