active-directory-acl-abuse

Warn

Audited by Gen Agent Trust Hub on Nov 16, 2026

Risk Level: MEDIUMPRIVILEGE_ESCALATIONCOMMAND_EXECUTIONPERSISTENCEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [PRIVILEGE_ESCALATION]: The skill provides detailed workflows for escalating from standard domain users to Domain Admin by exploiting misconfigured Access Control Entries (ACEs) such as GenericAll and WriteDACL.
  • [COMMAND_EXECUTION]: Instructions include numerous shell and PowerShell command-line templates for powerful security tools aimed at directory manipulation and information gathering.
  • [PERSISTENCE]: Technical methods are provided for long-term domain access, including the abuse of Group Policy Objects (GPOs) to deploy scheduled tasks and the use of Shadow Credentials via the msDS-KeyCredentialLink attribute.
  • [DATA_EXFILTRATION]: Explicit instructions for credential harvesting through DCSync attacks, LAPS password reading, and Targeted Kerberoasting are included.
  • [INDIRECT_PROMPT_INJECTION]: The skill operates on data ingested from Active Directory environments (e.g., BloodHound JSON results), creating a surface where malicious directory data could potentially influence agent behavior.
  • Ingestion points: BloodHound/Neo4j query results processed within the agent context via BLOODHOUND_PATHS.md.
  • Boundary markers: Missing delimiters to distinguish between instructions and untrusted data results.
  • Capability inventory: Extensive command execution (net user, Set-DomainObject), directory attribute modification (msDS-KeyCredentialLink), and network-based dumping tools (secretsdump).
  • Sanitization: No evidence of output sanitization or input validation for the ingested directory data.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 16, 2026, 12:04 PM