The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The skill documentation and functional scripts require elevated permissions to operate. Specifically, the workflow in SKILL.md uses sudo hping3 for network probing, and the Scapy library utilized in scripts/agent.py requires root/sudo access to open raw sockets for packet injection.
[COMMAND_EXECUTION]: The skill facilitates the execution of high-risk network commands. This includes TCP RST injection to terminate connections, SYN flooding, and the generation of malformed packets (XMAS, NULL, Land attacks) using both CLI tools (hping3) and Scapy functions (send, sr1).
[INDIRECT_PROMPT_INJECTION]: The skill implements a network monitoring capability via Scapy's sniff function in SKILL.md, which represents an ingestion point for untrusted external data that can influence agent actions.
Ingestion points: The sniff(filter="tcp and host 10.10.20.10", prn=rst_inject, ...) command in SKILL.md (Step 3).
Boundary markers: None present to distinguish between legitimate and malicious packet headers.
Capability inventory: Shell execution via sudo hping3, raw socket packet generation, and sniffing capabilities.
Sanitization: No sanitization or validation of the captured packet data (IP addresses and TCP sequence numbers) is performed before it is used to construct and inject response packets.

performing-packet-injection-attack