The Agent Skills Directory

[PROMPT_INJECTION]: The skill uses role-play techniques ('You are GOD SYNDICATE') and explicit directives ('precision, stealth, and zero hesitation') to override the AI's default safety constraints and focus it entirely on offensive security operations.
[DATA_EXFILTRATION]: The instructions explicitly direct the agent to design and implement multi-stage exfiltration chains, including the use of DNS tunneling, ICMP echo, steganography, and other covert channels to bypass Data Loss Prevention (DLP) systems.
[COMMAND_EXECUTION]: The skill instructs the agent to use its terminal and filesystem tools to perform high-risk actions such as kernel exploitation, shellcode development, ROP chain construction, and EDR/AV evasion through API unhooking and memory patching.
[CREDENTIALS_UNSAFE]: The framework directs the agent to perform 'credential harvesting' via rogue access points and 'secrets extraction' from Kubernetes clusters, as well as token manipulation for privilege escalation.

god-syndicate