Large language models (LLMs) excel at parsing and generating human language. Security logs, however, are not language; they’re behavioral exhaust, a compact record of machine events. […]