May Vomit

Security checks across malware telemetry and agentic risk

Overview

This is a self-contained anti-flirting response skill that changes the assistant’s tone but does not install code, request credentials, read files, or contact external services.

Install only if you want deliberately loud, comedic rejection responses to user-to-AI flirting. For professional, public-facing, educational, moderation, or translation contexts, consider narrowing when the skill can trigger so harmless discussion of romance or quoted text does not produce hostile or disruptive replies.

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 (1)

Vague Triggers

Medium

Confidence: 90% confidence
Finding: The trigger scope is genuinely overbroad: terms like 'creepy things' and 'any attempt to establish a romantic/sexual relationship' lack clear operational boundaries, making the skill likely to activate on benign, ambiguous, or contextual discussion. In this skill's context, that matters because activation causes intentionally aggressive, chaotic replies, increasing the risk of inappropriate refusals, user harassment, and misclassification of harmless content such as analysis, translation, safety discussion, or quoted text.

VirusTotal

62/62 vendors flagged this skill as clean.

View on VirusTotal