The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The skill explicitly instructs the agent to execute network scanning commands with administrative privileges using sudo. Examples include OS fingerprinting (sudo nmap -O), aggressive scanning (sudo nmap -A), and raw socket access for ARP discovery (sudo arp-scan-rs -l). This provides a path for the agent to acquire elevated permissions on the host system beyond what is strictly necessary for basic tasks.
[INDIRECT_PROMPT_INJECTION]: The 'Agentic Optimizations' section contains shell command templates that interpolate unvalidated variables like $IP and $PORTS (e.g., rustscan -a $IP -p $PORTS). This creates a vulnerability where malicious input data could lead to arbitrary command injection.
Ingestion points: Variable interpolation within shell commands in the 'Agentic Optimizations' section of SKILL.md.
Boundary markers: Absent. The skill lacks delimiters or instructions to treat inputs as literal data.
Capability inventory: High-privilege tool execution via the Bash tool (nmap, rustscan, arp-scan-rs).
Sanitization: No input validation or escaping logic is implemented for the provided variables.

network-discovery