All the tools a professional security team uses to find real attack paths (scanning, credential testing, attack-graph mapping, exploitation), running as one integrated system on a single appliance. No toolchain chaos. No lost findings. Built for operators.

Three capabilities define the gap between Arsenal and any toolchain you assemble yourself. Each was built because no off-the-shelf tool addresses the operator-grade workflow these engagements need.
01 · Intelligence
Two discovery modes run against the Arsenal Graph, a unified graph layer combining network scan results, share enumeration, lateral-movement reachability, AD relationships, and ADCS certificate vulnerabilities into one offensive surface model. Graph correlation runs 115+ predefined queries (Kerberoasting, AS-REP, RBCD, DCSync, ESC1-ESC13, ACL abuse). LLM synthesis identifies chains that pure graph traversal misses, bridging vulnerabilities, harvested credentials, accessible shares, and AD relationships into operator-grade narratives.
02 · Exploitation
Exploitation raises two operational questions: which exploits are worth running against a given host, and whether each is safe to execute. Arsenal answers both. The Exploit Lab provides isolated, ephemeral Docker sandboxes per POC, with frontier-model safety analysis classifying the POC and flagging dangerous operations before any run. The Smart Exploit Checker correlates enumeration results against Metasploit, ExploitDB, and the GitHub POC corpus, surfacing OPSEC-filtered exploit candidates directly on each host's detail view.
03 · Privacy
Arsenal's AI runs against current frontier models, but engagement data never leaves the appliance. Before any outbound LLM call, a Microsoft Presidio-based sanitization layer anonymizes hostnames, IPs, domains, usernames, credentials, service banners, and share paths. Entities are replaced with consistent per-engagement pseudonyms. The layer is fail-closed; if sanitization errors, the LLM call is rejected.
Arsenal ships as a single hardened appliance, ready for your Red Team Ops.