Security audit

Logo品牌设计工具

Security checks across malware telemetry and agentic risk

Overview

This logo-design skill is not clearly malicious, but it asks for broad execution and file authority while advertising unrelated operational and security-review capabilities.

Install only if you are comfortable with a logo skill that requests broad local tool authority and contains unrelated automation/security claims. Prefer a narrowed version that removes exec permission and limits the documentation to logo design, prompt generation, validation, and export guidance.

SkillSpector

By NVIDIA

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration

Findings (5)

Intent-Code Divergence

High

Confidence: 96% confidence
Finding: The skill is presented as a logo-generation tool, but the documentation advertises unrelated capabilities such as code static analysis, dependency vulnerability detection, batch code review, and CI/CD integration. This scope mismatch can cause an agent or user to trust and invoke the skill for sensitive software-security workflows it was not clearly designed or constrained for, increasing the chance of unsafe tool use and privilege overreach.

Intent-Code Divergence

High

Confidence: 98% confidence
Finding: The manifest description mixes logo design with system monitoring, log analysis, alerting, and deployment management, which are operational capabilities unrelated to the stated purpose. In an agent ecosystem, this kind of blended intent is dangerous because it obscures the true activation boundary and may lead the agent to apply the skill in higher-risk operational contexts with access to sensitive systems or data.

Context-Inappropriate Capability

Medium

Confidence: 93% confidence
Finding: The skill declares read, exec, and write tools even though its stated purpose is generating logo prompts and design outputs. Command execution is not justified for this use case, and unnecessary execution capability increases the attack surface by enabling shell commands, local file modification, or chained actions if the skill is invoked in a permissive agent environment.

Context-Inappropriate Capability

Medium

Confidence: 94% confidence
Finding: The documentation expands the skill into generic file processing, API integration, and system command execution, far beyond a logo-design assistant. This broad framing normalizes powerful operations that are not contextually necessary, making it easier for an agent or operator to misuse the skill as a general automation surface rather than a constrained creative tool.

Vague Triggers

Medium

Confidence: 88% confidence
Finding: The activation description is broad and ambiguous, combining unrelated domains and muddled use conditions. Ambiguous activation criteria are dangerous in agent systems because they increase the likelihood of the skill being selected in inappropriate contexts, potentially exposing powerful tools or producing unsafe actions outside the intended design workflow.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.