ui-ux-pro-max
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOWCOMMAND_EXECUTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [Command Execution] (LOW): The skill workflow involves executing shell commands with user-controlled keywords. Evidence:
SKILL.mdtemplatepython3 .../search.py "<keyword>". Risk: Potential command injection if keywords are not properly escaped by the agent platform. - [Privilege Escalation] (INFO): Documentation contains
sudocommands for environment setup. Evidence:SKILL.mdlistssudo apt update && sudo apt install python3. - [External Downloads] (LOW): The skill documentation recommends downloading the Python runtime from OS-specific package managers. Evidence:
SKILL.mdcontainsbrew install,apt install, andwinget installcommands. - [Indirect Prompt Injection] (LOW): User input is interpolated into shell commands which influence agent reasoning. 1. Ingestion points: User requests used as keywords in
SKILL.md. 2. Boundary markers: Absent in command templates. 3. Capability inventory: Read-only file access viascripts/search.py. 4. Sanitization:BM25.tokenizeincore.pyremoves punctuation but does not protect against shell breakout.
Audit Metadata