ui-ux-pro-max

Fail

Audited by Gen Agent Trust Hub on Feb 15, 2026

Risk Level: HIGHCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • COMMAND_EXECUTION (HIGH): The skill instructions in SKILL.md direct the agent to execute shell commands using user-provided keywords (e.g., python3 .../search.py "<keyword>"). This pattern is vulnerable to command injection where malicious users can execute arbitrary code on the host system using shell metacharacters like $() or backticks.
  • COMMAND_EXECUTION (HIGH): The skill requests privilege escalation by instructing the agent to use sudo for software installation on Linux systems (sudo apt install python3), which grants the agent root access to the host.
  • PROMPT_INJECTION (HIGH): The skill is susceptible to indirect prompt injection through user-controlled input. Ingestion point: Step 1 requirements extraction from user requests. Boundary markers: Absent. Capability inventory: Shell execution via python3 script calls. Sanitization: None detected for the keyword interpolation in the shell command.
  • EXTERNAL_DOWNLOADS (LOW): Recommends installing Python via trusted package managers (brew, apt, winget). Under [TRUST-SCOPE-RULE], these are considered trusted sources, but automated agent-led software installation is generally a poor security practice.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 15, 2026, 09:08 PM