is just a symbolic link to that).
interactively. The challenge with either of these approaches is detecting and handling errors. I think most teams that want to use osquery as a library really just want its abstraction of OS APIs, but don't really need its SQL abstractions. Unfortunately there's no clean partitioning of those two things in the code. If there was, maybe all of the OS APIs could be a library used by both the osquery agent and other teams.