windows-privilege-escalation

Fail

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: HIGHPRIVILEGE_ESCALATIONCREDENTIALS_UNSAFEPERSISTENCECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [PRIVILEGE_ESCALATION]: The skill provides comprehensive instructions for escalating privileges on Windows systems through various vectors.
  • Includes multiple techniques for User Account Control (UAC) bypass using binaries like fodhelper.exe, eventvwr.exe, and WSReset.exe.
  • Details the abuse of token privileges such as SeImpersonatePrivilege, SeDebugPrivilege, and SeBackupPrivilege using the 'Potato' family of exploits.
  • Provides methods for exploiting service misconfigurations (unquoted paths, weak permissions) and scheduled task abuse.
  • [CREDENTIALS_UNSAFE]: Instructions are included to directly access and extract sensitive authentication data.
  • Commands for dumping the memory of the Local Security Authority Subsystem Service (LSASS) via procdump -ma lsass.exe.
  • Methods for extracting sensitive registry hives (SAM, SYSTEM, SECURITY) using reg save to retrieve local account secrets.
  • Enumeration of stored credentials via cmdkey /list and direct access to AppData credential storage paths.
  • [PERSISTENCE]: The skill outlines techniques to maintain access across sessions.
  • Explicit command to create a new user and add them to the local administrators group: net user hacker P@ss123 /add && net localgroup administrators hacker /add.
  • Instructions for hijacking registry autorun keys (HKLM\...\Run) to ensure malicious code executes upon system startup.
  • Guidance on creating malicious scheduled tasks to run as the SYSTEM user.
  • [COMMAND_EXECUTION]: Extensive use of shell commands and system utilities to alter system state.
  • Reconfiguring services using sc config to point to malicious binaries.
  • Modifying file and directory permissions using icacls and takeown.
  • Manipulating the Windows registry to change execution policies and environment variables.
  • [DYNAMIC_EXECUTION]: Employs techniques to load or execute code dynamically at runtime.
  • Detailed instructions for DLL hijacking by placing malicious files in high-priority search directories or using 'Mock Trusted Directory' tricks.
  • Spawning system-level shells by migrating to existing processes or using token impersonation tools.
  • Instructions for generating and executing malicious MSI packages via msiexec.
  • [REMOTE_CODE_EXECUTION]: Provides patterns for executing code from remote sources.
  • The CMSTP UAC bypass method includes a pattern to execute a remote Scriptlet (.sct) payload via a URL.
  • [INDIRECT_PROMPT_INJECTION]: The skill presents an attack surface where untrusted system data is processed to determine next steps.
  • Ingestion points: Ingests output from system enumeration commands (systeminfo, whoami /all, wmic, sc query, schtasks) across all files.
  • Boundary markers: None identified; system output is processed directly to guide exploitation.
  • Capability inventory: Significant capabilities including administrative command execution, registry modification, file system manipulation, and network operations (PrintSpoofer64.exe, GodPotato.exe, Akagi64.exe).
  • Sanitization: No evidence of sanitization or validation of the system output before the agent acts upon it.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Sep 15, 2026, 02:13 PM
Security Audit — agent-trust-hub — windows-privilege-escalation