Suggestion from Theodore:

Suggested langauge improvement from the reproducing bugs section --- tell people to try building with clang and the relevant santizer, and then try running the fuzzing target against the test case. If it doesn't reproduce, or if they don't have access to a new-enough version of clang, then they can fall back to using the Docker environment. For most people who don't have deep experience with Docker, that's actually going to be simpler than trying to use infra.py and trying to wrestle with docker. Also, if they learn how to build the sanitizer into their application, and exposing them to the clang docs, they might be able to find more bugs on their own (e.g., by running their regression test suite with various sanitizers enabled).