Here are the collections I know of, most recently updated first:

2021 Schemepunk (https://github.com/ar-nelson/schemepunk)
2021 Yuni (https://github.com/okuoku/yuni)
2019 Xitomatl (https://github.com/DerickEddington/xitomatl)
2019 Wak (https://gitlab.com/wak) -- old code with some porting patches
2017 Spheres (https://github.com/alvatar/spheres)
2015 Klio (https://github.com/mbenelli/klio)
2015 Spells (https://github.com/rotty/spells)
2011 Nausicaa (https://github.com/okuoku/nausicaa)

For our purposes, Schemepunk and Yuni are the only ones that are
actively being worked on.

Adam is on board with adapting Schemepunk for wider use.

Yuni is more of an R7RS portability layer than a library collection. I
was able to reach its author, and he says it's being used for a
commercial product and hence needs to stay as its own entity with
somewhat idiosyncratic requirements.

Hence, it would seem the best start for a Scheme Live is to take
Schemepunk and many SRFIs (esp. the R7RS-large ones), and merge code
from the other collections above until we have all the features they
cover, changing the interfaces as needed so the result is cohesive.

Opinions?