limtdesign

• Outcome: Pass
• Latest audit: May 29, 2026, 8:13 PM
• Version:
  1.0.0

Overview

This is a markdown-only Chinese helper for creating visual AI image prompts, with no code execution, persistence, credential access, or hidden system behavior found.

Safe to install for Chinese-language visual prompt generation. Users who do not read Chinese may find the skill hard to use, and users uploading product, model, scene, or style images should only provide files they are comfortable sending to the downstream image-generation service.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Intent-Code Divergence

Low

• Confidence: 94% confidence
• Finding: The instruction to treat a character reference sheet as a "product image" is a real semantic mismatch that can cause the agent or downstream tooling to invoke the wrong reference-image pipeline. In a prompt-generation skill, this can lead to incorrect data handling, broken policy enforcement, or unintended preservation/transformation behavior for user-supplied images, making the skill less predictable and potentially bypassing scene-specific safeguards.

Vague Triggers

High

• Confidence: 93% confidence
• Finding: The skill’s trigger criteria are extremely broad, covering generic requests like making a poster, creating visuals, or generating prompts. In an agent environment, this can cause the skill to activate for many unrelated or only loosely related requests, creating routing hijack risk, reducing user intent fidelity, and potentially overriding more appropriate specialized skills.

Natural-Language Policy Violations

Medium

• Confidence: 76% confidence
• Finding: The skill metadata and content are written to enforce Chinese-language behavior without indicating that output should follow the user’s preferred language. This can create instruction-level mismatch, degrade usability, and in multi-skill systems may cause unintended behavior or concealment of outputs from users expecting another language, though it is not a direct code-execution or data-exfiltration risk.

Natural-Language Policy Violations

Medium

• Confidence: 94% confidence
• Finding: The skill content is entirely written to operate in Chinese and does not indicate that the assistant should adapt to the user's preferred language or ask for language choice. This can cause unintended behavior, reduce usability, and create prompt-steering risk where user instructions in another language are overridden by the skill's implicit language requirement.

Natural-Language Policy Violations

Medium

• Confidence: 90% confidence
• Finding: The file is entirely written in Chinese and gives no indication that the skill can adapt to the user's preferred language. In a multilingual agent environment, this can cause misunderstanding of instructions, degraded usability, and incorrect outputs for users who do not read Chinese, though it does not directly create a code-execution or data-exfiltration risk.

VirusTotal

64/64 vendors flagged this skill as clean.