YARA vs Kubernetes

Every release in this window is memory safety work on the parsers. The July batch bounds the rule table index in two opcodes, the tilde stream row count in the dotnet module and the repeat stack depth in the regex fiber sync, and fixes a leak in rule stream loading. Earlier releases closed a heap overflow triggered by hand-crafted compiled rules, infinite loops on corrupt PE resource directories, and an integer overflow in ELF parsing. The three most recent tags were published within thirteen minutes of each other.

YARA is being maintained as an input parser under adversarial pressure rather than developed as a language. The last release to add anything — 4.5.0, with unreferenced string rules, strict escape warnings and a slow-rule callback — is nearly two years back, and everything since has been bounding a value someone found a way to overflow. Even the scan limit change was a revert to a prior default.

Expect the pattern to hold: batched patch releases whose contents are bounds checks in the PE, dotnet and regex paths, since that is where every recent finding has landed. Nothing in these entries suggests new language or module capability is queued.