DDoS Attack Testing

Warn

Audited by Gen Agent Trust Hub on Nov 6, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill downloads third-party scripts from external, unverified repositories and executes them locally.
  • Evidence: Downloads slowloris.py from github.com/gkbrk/slowloris and goldeneye.py from github.com/jseidl/GoldenEye via git clone, followed by execution with python.
  • [DYNAMIC_EXECUTION]: External scripts are downloaded and executed at runtime.
  • Evidence: python slowloris.py 192.168.1.107 -p 80 -s 500 and python goldeneye.py http://192.168.1.107 -w 50 -s 500.
  • [PRIVILEGE_ESCALATION]: The skill requires administrative privileges for several operations.
  • Evidence: Usage of sudo apt-get install for tool setup, sudo gedit for modifying system configuration files like /etc/snort/rules/local.rules, and sudo snort for running the IDS.
  • [COMMAND_EXECUTION]: The skill executes high-impact network tools that can cause service disruption.
  • Evidence: Extensive use of hping3 with the --flood flag, msfconsole for SYN flooding, and GUI tools like LOIC/HOIC for traffic generation.
  • [INDIRECT_PROMPT_INJECTION]: The skill presents an attack surface where user-supplied input is directly interpolated into shell commands without validation or boundary markers.
  • Ingestion points: Target IP addresses and URLs provided by the user in SKILL.md (e.g., Phase 2, Phase 9).
  • Boundary markers: None identified.
  • Capability inventory: hping3, msfconsole, python (executing scripts), snort, wireshark via shell commands.
  • Sanitization: None identified; user input flows directly into tool arguments.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 6, 2026, 03:30 PM