Security audit

Ui Ux Promax V2 Free

Security checks across malware telemetry and agentic risk

Overview

The skill is a design-guidance lookup skill, but it asks for command execution and repeatedly tells agents to run a missing relative Python script, which needs review before installation.

Review before installing. Only use this skill if you can verify the intended scripts/search.py and data files are present and trusted. Do not let the agent run relative python3 scripts/search.py commands from arbitrary project directories, and handle any Python or sudo package installation steps yourself.

SkillSpector

By NVIDIA

Vulnerability Patterns

Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection

Findings (7)

Description-Behavior Mismatch

Medium

Confidence: 88% confidence
Finding: The skill is presented as a local, data-file-backed design query tool, yet its error handling normalizes network diagnostics and connectivity troubleshooting. That mismatch can expand the operational scope beyond what users expect and may induce an agent to run unnecessary network-related commands, increasing the attack surface for a skill that should remain local-only.

Context-Inappropriate Capability

Medium

Confidence: 93% confidence
Finding: The skill enables exec despite being described as a design-guideline lookup database. Granting command execution for a largely informational skill increases the risk that an agent will run local commands unnecessarily, which is dangerous because even simple setup or lookup flows can become a vector for filesystem, process, or environment interaction.

Intent-Code Divergence

Medium

Confidence: 78% confidence
Finding: The documentation says the free version does not support persistence, but another section says operations support create/query/export. This inconsistency can mislead an agent or user about whether data creation or export side effects are expected, which matters in a tool executed with command capability.

Vague Triggers

Medium

Confidence: 84% confidence
Finding: The trigger condition is very broad and overlaps with common design, branding, and creative requests, increasing the chance the skill activates in contexts where the user did not specifically ask for it. Overbroad auto-invocation is risky because this skill also encourages command execution, so routine prompts could unnecessarily escalate into tool use.

Sudo/Root Execution

Medium

Category: Privilege Escalation
Content: brew install python3 # Ubuntu/Debian sudo apt update && sudo apt install python3 # Windows winget install Python.Python.3.12
Confidence: 95% confidence
Finding: The repeated root-execution finding is valid for the same line because the skill explicitly instructs use of sudo for package installation. Any skill that combines broad exec capability with privileged command examples raises the chance of harmful host-level modification if the instructions are followed or operationalized by an agent.

Sudo/Root Execution

Medium

Category: Privilege Escalation
Content: brew install python3 # Ubuntu/Debian sudo apt update && sudo apt install python3 # Windows winget install Python.Python.3.12
Confidence: 95% confidence
Finding: The repeated root-execution finding is valid for the same line because the skill explicitly instructs use of sudo for package installation. Any skill that combines broad exec capability with privileged command examples raises the chance of harmful host-level modification if the instructions are followed or operationalized by an agent.

Chaining Abuse

High

Category: Tool Misuse
Content: brew install python3 # Ubuntu/Debian sudo apt update && sudo apt install python3 # Windows winget install Python.Python.3.12
Confidence: 94% confidence
Finding: The chained command 'sudo apt update && sudo apt install python3' compounds risk by performing multiple privileged actions in one step. Chaining reduces opportunities for review, makes accidental execution more impactful, and in an agent setting can normalize opaque multi-step system modification for a skill whose purpose is unrelated to administrative package management.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.