AITU Final multi-AI manager. Use when one AI is coordinating multiple Codex/Claude workers, workers ask what to do next, report blocked/done/high-value findings, or the user wants one manager to assign next tasks without manually maintaining a long dispatcher prompt.
Pivot/Tunneling autonomous skill: automatic segment detection from a single pivot IP, automatic SSH credential attempts, tunnel establishment (ligolo/ssh/chisel), internal network host/port discovery, and automatic MCP reflection (add_host/update_host/import_scan_data).
Automatic Bug Bounty report generation. Collects MCP data from report_id, host_ip:port, or host_ip:port bug_type, generates Description, registers/updates via web app, and saves attachments.
Automatic Risk report generation. Auto-infers the target from a Report ID, risk name, or the current MCP attack chain context to generate a Description, register in the web app, and save attachments.
AITU Final autonomous orchestrator. Use when the user says continue, keep going, do the next thing, or gives no target context. Read MCP state, lock onto one active risk objective, and dispatch scanning, web, AD, pivot, SCADA, and reporting without asking the user to choose a host or risk.
AD autonomous attack (token-optimized): enum -> BH/Kerberos -> spray -> privesc -> DCSync. If no arguments are given, automatically selects the next domain/DC target from MCP and proceeds.
SCADA/ICS attack (token-optimized): protocol identification -> per-protocol enumeration/verification -> value manipulation -> evidence capture. If no arguments are given, automatically selects a Hackcity/OT target from MCP.
Web/DMZ attack (token-optimized): source collection -> code analysis -> vulnerability verification -> evidence/report. Even without a target, automatically selects a reachable web target from MCP and continues.