Password Spray with Adaptive Lockout Evasion Timing

Detects password spray activity that distributes attempts across accounts at a cadence deliberately calibrated to stay below lockout thresholds. Human operators spray at a fixed interval; autonomous AI agents adaptively tune their attempt rate based on observed lockout responses. The diagnostic is a sustained, evenly paced pattern of single-attempt failures across many accounts — one attempt per account, then a pause just longer than the observation window, then the next password across the same set. ARTEX planner agents track lockout policy details discovered during recon and configure worker agents to spray below the lockout threshold. The uniform cadence and breadth (many accounts, one attempt each, repeated) distinguishes this from legitimate failed logins which cluster around specific accounts. Requires Windows Security Event 4625 (failed logon). Lower the account threshold for environments with fewer interactive users. BQL (Bloo Query Language): stream=authentication where action='LOGIN' and status='FAILED' and logon_type in ('3', '10') and not user like '%$' | groupby source_ip, user | having count > 15