I haven't looked at the Lemmy API, but generally speaking hard or not is subjective. Try it, get your hands dirty. Take a stab at testing how it works. Set some very basic targets. Split it up into tasks. If one task proves more complicated than expected, split it up into smaller tasks. If you get stuck, move to another task. You can always get back to those later. Things fall in place with experience.
Originally communication on the web was one directional, server to client. Web 2.0 meant active web and bidirectional communication. Hence, web 3.0 is a threesome.
There is probably a neat wget oneliner that could crawl everything on the open web. The real challenge is how to index all the information. That might be a neat Perl oneliner.
DDG, but ever so often I have to use bangs to Google or some dedicated sites. Been trying out my own instance of the SearchX meta search engine but honestly it's not that much of a difference except it lags. I've been using ChatGPT way more for direct questions instead of using search words and sifting through the results, with the risk of hallucination so still need to double check on important stuff. Copilot sort of works but I don't know why I'm not comfortable with it. Too bad Gemeni seems to be a dud so far. I'd love to see a FOSS language model that can be tweaked for personal interests and custom commands for external APIs and that gives references to the answers.
FairEmail. I got fed up with the Gmail app injecting ads despite I pay for storage and services. FairEmail is not beautiful by modern standards but with some tweaking I can live with it, no problem.
The irresponsible pitbull owners mindset right there.