I have a similar feeling towards Kanji Portraits; it differs from many other sources, and seems to lean towards idiomatic compound etymologies, even when there is a straightforward and commonly accepted phono-semantic etymology (not that the two are mutually exclusive).
That being said, sometimes it does provide some very interesting analysis, pointing out how a particular character has changed over time.
Personally, I like to check Wiktionary first, and then compare the results with Kanji Portraits and yellowbridge, to see if they differ. I also like to look up some characters myself on Chinese Etymology (quick and easy) and Sinica (slower and harder to use, as it is in Chinese, but it provides additionaö details that helps you track down the actual source insceiption, for example by taking the collection ID of an oracle bone and entering it here).
Also, while not a free resource, I really like the book The Complete Guide to Japanese Kanji by Kenneth Henshall; he cites sources and compares historical variants, and also theorizes about why each character has the components it does, which can often be more helpful than Wiktionary’s often quite brief etymologies. His analyses aren’t necessarily guaranteed to be correct, but it’s nice to be able to compare my own armchair analyses with that of an actual researcher.