Khazix Writer 卡兹克写作

Security checks across malware telemetry and agentic risk

Overview

This is a Chinese long-form writing style skill that only gives drafting guidance and does not include code, credential access, persistence, or data-moving behavior.

Install this if you want drafts shaped in the Khazix public-account article style. For generic or sensitive writing, explicitly ask for a neutral tone or your required formatting, and review any external GitHub materials before asking an agent to use them.

SkillSpector

By NVIDIA

Vulnerability Patterns

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

Findings (2)

Vague Triggers

Medium

Confidence: 91% confidence
Finding: The activation text is broad enough to trigger on common requests like 「帮我写篇文章」 or 「写稿子」 without clear limits on topic, risk level, or required user consent to adopt this persona/style. In an agent environment, overly broad routing can cause this skill to take over general writing tasks unexpectedly, overriding user intent or applying hidden formatting and behavioral constraints across unrelated content.

Natural-Language Policy Violations

Low

Confidence: 86% confidence
Finding: The skill hard-codes stylistic rules like banning certain punctuation and requiring specific quotation marks, but does not indicate that these conventions are optional or contingent on user preference. This can silently override user formatting requirements, editorial standards, accessibility needs, or downstream publishing constraints, creating integrity and usability issues rather than direct code-execution risk.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal