The tools work by detecting two anti-patterns the researchers pinpointed after analyzing thousands of malicious code instances in repositories and packages: obfuscated / unreadable source code, and ...