Kliper is a relatively modern design, which, while admittedly not produced in metal, used much more benign components for RCS. And Kliper also doesn't have a separate service module.
One can always say "but Klipper is a paper spacecraft". Of course, nothing is perfect. Buran can be brought as an example, and arguments against that could be broght up - "oh, it's too old". The point still stands - it's possible to not use toxic components, it's desirable - and it's not done. In my opinion, performance arguments - for a relatively low delta-V system - aren't convincing. After all, Crew Dragon is much bigger and newer craft that Soyuz.
Paper spacecraft can do anything. Buran was designed for the same mission lengths as the Shuttle, on the order of a few weeks at most, not half a year or more. For long mission durations, storable hypergolics are the way to go for now. We don't have good flight time on anything else, and we need that for something we're putting crew on.
> For long mission durations, storable hypergolics are the way to go for now.
Soyuz is a counterexample which flies for decades. Lockheed's engineers working on orbital fuel storage (Advanced Cryogenic Evolved Stage) will also point that it's not a law of physics.
One can always say "but Klipper is a paper spacecraft". Of course, nothing is perfect. Buran can be brought as an example, and arguments against that could be broght up - "oh, it's too old". The point still stands - it's possible to not use toxic components, it's desirable - and it's not done. In my opinion, performance arguments - for a relatively low delta-V system - aren't convincing. After all, Crew Dragon is much bigger and newer craft that Soyuz.