One option would be to host a public text file (on Github, even) that describes various services, their password requirements, domain aliases, and have the script tack on the approriate strings based on the public text file. Others could even use the same script and submit pull requests describing new services they encounter.
My main objective is to eliminate personal databases.
Okay, so say I submit a pull request for a phishing domain pretending to be an alias of something, and months/years later trick you into visiting it. Seems risky to me.
This would only be risky if I got phished into visiting your phishy domain AND was stupid enough to consciously want to enter a password into it, in which case the risks are the same as if I used a human-memorized password.
My password "manager" doesn't automatically fill in passwords until I consciously ask it to.
But some amount of human oversight over those pull requests would be in order.
My main objective is to eliminate personal databases.