The Agent Skills Directory

[OBFUSCATION]: The skill uses string manipulation in its documentation and payload examples, such as syst[e]m instead of system, to prevent security software from flagging the skill files as malicious.
[DATA_EXFILTRATION]: The instructions provide methods to access and read sensitive files on remote targets, including /etc/passwd, through various service-specific commands.
[PERSISTENCE]: The skill includes techniques for establishing persistent access to remote systems, specifically by using Redis to write a public SSH key into the root user's authorized_keys file.
[DYNAMIC_EXECUTION]: The skill facilitates the creation and execution of dynamic payloads, including PHP webshells and database commands (xp_cmdshell, sys_exec, COPY FROM PROGRAM) designed to run arbitrary system commands.
[COMMAND_EXECUTION]: The skill makes extensive use of powerful network tools (nmap, masscan, smbclient, ldapsearch) to interact with and exploit remote infrastructure.
[EXTERNAL_DOWNLOADS]: The skill references and directs the user to external security documentation and payload repositories on GitHub, such as HackTricks and PayloadsAllTheThings.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external, potentially untrusted network services, creating a surface for indirect prompt injection.
Ingestion points: Service banners and remote data retrieved via nmap, nc, smbclient, and other tools.
Boundary markers: No explicit markers or instructions are provided to delimit or ignore instructions embedded in data from remote services.
Capability inventory: The skill possesses extensive network access, file writing capabilities, and command execution features.
Sanitization: There is no evidence of filtering or sanitization of data retrieved from external services before it is processed by the agent.

enumerating-network-services