Show HN: An AI-native, offline malware scanner for Linux (it's free) (github.com)

SemanticsAV is an open-source, AI-native malware scanner for Linux that emphasizes offline, semantics-based detection rather than signatures or sandbox heuristics. The project ships as three pieces — an on-device SemanticsAV SDK (lightweight AI engine for instant, offline verdicts), an optional SemanticsAV Intelligence cloud API (for forensic context, campaign attribution and “genetic” neighborhood mapping), and the SemanticsAV CLI/gateway (daemon, HTTP/Unix-socket API, and automation-friendly tools). Because the SDK reasons about code meaning and contextual patterns, the authors claim improved zero-day and polymorphic detection that evades traditional pattern-based evasion. Technically it’s production-ready: multi-threaded scanning, CI/CD integration, JSON/HTML/Markdown reports, daemon mode with REST and a zero-copy binary socket protocol, and deterministic, minimal analytical payloads for privacy-preserving cloud analysis (original files never transmitted). Supported formats today are PE and ELF on x86_64 and aarch64 (glibc/libstdc++ compatibility listed); models are updated periodically and both community and commercial editions receive models (commercial gets prioritized/NDA-trained samples). For practitioners this promises instant, air-gapped scanning at scale with optional rapid cloud attribution, while preserving auditability via MIT-licensed networking wrappers and open-source CLI/daemon code.