I'm a little torn between
auto is too generic.
On the other hand,
javaagent is maybe too restrictive. We have already had discussions about re-using the bytecode instrumentation outside of a javaagent (e.g. bootstrapping during
main() for aws lambda, and compile-time instrumentation for android).
OTOH, it makes sense to name things for their primary use cases (in our case, using
javaagent instead of
auto). I think it would be ok if we reuse those artifacts for other use cases (e.g. using
javaagent artifacts for compile time android instrumentation).