Security audit

文献自动解析

Security checks across malware telemetry and agentic risk

Overview

This appears to be a narrow Chinese-language literature analysis skill with prompt-scoping quality issues but no evidence of data theft, persistence, code execution, or destructive behavior.

Install if you specifically want Chinese-language literature or text analysis. Be aware it may activate too broadly and may answer in Chinese even when that is not what you wanted; avoid using it for unrelated prompts where strict language or formatting could get in the way.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (2)

Vague Triggers

Medium

Confidence: 95% confidence
Finding: The skill instructs that any user-sent text should be treated as literature content, creating an overly broad activation scope. This can cause the skill to intercept unrelated prompts and override user intent, which is risky in agent environments because it may mis-handle arbitrary inputs and suppress safer or more appropriate behaviors.

Natural-Language Policy Violations

Medium

Confidence: 82% confidence
Finding: The skill hard-codes Chinese-only output and forbids any extra explanation or clarification, regardless of the user's language or context. While not directly enabling code execution or data exfiltration, this can cause unsafe misalignment with user expectations, reduce transparency, and make the skill easier to misuse in multilingual settings where accurate understanding matters.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.