Wirewolf is built to read captures of live malware traffic. Its design rules:
- Packets are parsed as data. Nothing from a capture is executed, opened with another program or exported to disk. (Wireshark's "Export Objects" is never invoked.)
- Parsing runs in a separate
ww workerprocess with a timeout; each frame is additionally wrapped in panic containment. Parsers are fuzzed with mutated real captures (crates/ww-parse/tests/mutation_fuzz.rs). - The application makes no network requests. Threat-intel feeds are fetched only when you run
scripts/update_feeds.pyyourself. - The Tauri window has core permissions only: no shell, filesystem or network plugins; the content security policy allows no remote origins.
scripts/fetch_mta.pynever downloads archives that contain samples or artefacts and extracts only members with a PCAP/PCAPNG (or answer PDF) signature.
Even so, analyse untrusted captures inside a virtual machine. Antivirus software on the host may quarantine downloaded captures; keep your corpus in a dedicated, excluded folder.
If you find a way for a crafted capture to crash the parser, hang it, write outside the case cache or execute code, please open a private security advisory on GitHub (Security → Report a vulnerability) with a minimal reproducer. Do not attach real malware samples; a synthetic or truncated capture is enough.