fluke is still experimental, but you can help by trying to build stuff with it.
The highest-impact work right now is probably picking up an h2spec
case and making it pass, or tackling some of the TODO: or FIXME:
in the code.
More test coverage is always welcome, it's tracked via llvm-cov, GitHub
actions will comment on PR with coverage changes, and you can see the
coverage for yourself with just cov, in the coverage/html directory.
Some more design needs to happen, especially re: configuration and introspection of running HTTP services, that'll happen when H1/H2 work relatively well.