ClawHub
Back to skill

Security audit

milady

Security checks across malware telemetry and agentic risk

Overview

This is a small style-writing skill with broad aesthetic steering but no code execution, data access, persistence, or hidden authority.

Install this if you specifically want Network Spirituality / Remilia / Milady aesthetic writing. Be aware that its trigger wording is broad, so it may influence some general art or creative-writing requests unless your agent lets you control when the skill is used.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • 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 (3)

Vague Triggers

Medium
Confidence
87% confidence
Finding
The top-level description is broad enough to trigger on many generic creative-writing or art requests, which can cause unintended activation and steer responses into a niche ideology/aesthetic the user did not request. In a style skill, this is primarily a scope/consent problem rather than direct code-execution risk, but it can still degrade alignment and user intent handling.

Vague Triggers

Medium
Confidence
90% confidence
Finding
The 'When to Use' list includes ambiguous categories like art descriptions, writing in a voice, and technological/spiritual themes without clear boundaries, increasing the chance the agent applies this persona to unrelated requests. Because the skill strongly frames content through a specific subcultural lens, accidental activation could override neutral assistance and introduce unsolicited ideological or stylistic steering.

Natural-Language Policy Violations

Low
Confidence
80% confidence
Finding
Directing the model to use bilingual elements 'when appropriate' can cause language mixing without explicit user consent, reducing clarity or accessibility for users who expect a single language. In this context the effect is limited to style drift rather than a severe security failure, but it still represents avoidable output steering beyond the user's request.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.