The Agent Skills Directory

[COMMAND_EXECUTION] (MEDIUM): The skill constructs and executes complex shell commands involving qemu-system-x86_64, pkill, and network diagnostic tools (nc, telnet). If parameters like ports or image paths are sourced from untrusted input, it could lead to command injection.
[PROMPT_INJECTION] (MEDIUM): Implements an Indirect Prompt Injection surface. The agent is instructed to read and grep output from the VM's serial console (telnet localhost PORT). If a malicious ISO is used, the VM output could contain instructions designed to manipulate the agent's logic.
Ingestion point: VM serial console output via telnet (SKILL.md).
Capability inventory: Execution of qemu-system, pkill, and shell loops (SKILL.md).
Sanitization: Minimal; uses grep for specific strings but does not sanitize or escape the overall data stream.
Boundary markers: None present to distinguish VM output from system instructions.

qemu-startup