The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts scripts/agent.py and scripts/process.py utilize the subprocess module to execute external tools including ldapsearch, powershell, and wevtutil, as well as components from the impacket library. These operations are necessary for interacting with Active Directory and local security logs as part of the skill's primary function.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from potentially untrusted sources such as LDAP search results and Windows Security Event files (.evtx).
Ingestion points: Service Principal Name (SPN) data from LDAP queries in agent.py and process.py, and event data from .evtx files in agent.py.
Boundary markers: The scripts do not use explicit delimiters to isolate untrusted data within internal processing.
Capability inventory: The skill has the capability to execute shell commands and write files to the system (e.g., report generation).
Sanitization: Ingested data is parsed and filtered for specific attributes (e.g., sAMAccountName, servicePrincipalName) using standard libraries like ldap3 or xml.etree.ElementTree.

performing-kerberoasting-attack