The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest user-provided data such as IP addresses, URLs, and account names which are then used as parameters for offensive security tools. This creates an attack surface for indirect prompt injection.
Ingestion points: User-supplied target identification (IP, URL, account) mentioned in the confirmation gate of SKILL.md.
Boundary markers: Includes a 'Mandatory confirmation gate' requiring the agent to ask for authorization, show exact commands, and wait for explicit user confirmation.
Capability inventory: The skill integrates several high-capability tools for AD enumeration, credential dumping, and identity exploitation like NetExec, BloodHound, and Mimikatz (SKILL.md).
Sanitization: No specific input sanitization or validation methods for the interpolated target data are described in the instruction files.
[COMMAND_EXECUTION]: The skill includes numerous bash command examples for security assessment tools such as Certipy, Impacket, Responder, and Coercer. These tools are capable of performing sensitive network operations and protocol manipulations.

windows-ad