[Environments] collect n level of depth following the human trajectory of pages visited

[josancamon19]

• Follow each link that follows to the same host for n depth
• Explore each link, cache it as part of the bundle and update sample

[josancamon19]

can we do an enrich_environment script, that takes a task, opens the capture replay, and navigates to every possible link increasing the environment bundle collected?

or rather launch amazon.com in a separate script, and have a script that collects a sandbox of every possible section navigated.

curious how much data would this be
and then run the human and agent over this sandbox, re-sharing sandbox per task, not necessarily using a new sandbox per task
either way start by collecting 1 sandbox fully, e.g. amazon.com, or apple.com

can we then generate tasks for this website and automate any interaction on that specific website?