pentest-skills-framework

Warn

Audited by Gen Agent Trust Hub on Jun 12, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the user to clone its core repository from an external source (github.com/crazyMarky/pentest-skills.git) and install several third-party security tools from GitHub (via go install) and PyPI (via pip3). These sources are not included in the primary trusted vendor list.
  • [COMMAND_EXECUTION]: The provided Python scripts (e.g., port_scanner.py, subdomain_enum.py, full_recon.py) utilize the subprocess module to execute various CLI tools like nmap, subfinder, dnsx, httpx, and ffuf using arguments derived from user input.
  • [REMOTE_CODE_EXECUTION]: The installation instructions involve downloading a repository and copying its content directly into the agent's skills directory (~/.claude/skills), which facilitates the execution of scripts downloaded from the internet.
  • [PROMPT_INJECTION]: The skill's 'Automated Analysis' feature processes output from external network scans (such as service banners, web page titles, and technology fingerprints). This represents a surface for indirect prompt injection where malicious data from a target system could attempt to influence the agent's behavior.
  • [PRIVILEGE_ESCALATION]: The documentation provides explicit commands for using sudo to install system packages and troubleshooting advice for running security tools with elevated permissions (e.g., sudo nmap -sS).
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jun 12, 2026, 08:51 PM
Security Audit — agent-trust-hub — pentest-skills-framework