Game Theory

Security checks across malware telemetry and agentic risk

Overview

This is a markdown-only game theory reference skill for crypto analysis, with some dual-use attack discussion but no code, credentials, persistence, or hidden execution.

Safe to install as a reference skill. Use the governance and MEV attack material for defensive analysis, protocol hardening, and risk review; do not use it to plan manipulation, exploit execution, or deceptive governance activity against real communities or protocols.

SkillSpector

By NVIDIA

Vulnerability Patterns

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
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (1)

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The referenced section explicitly gives operational guidance to attackers (e.g., how to accumulate quietly, time proposals, divide opposition, and extract incrementally) without nearby safety framing, defensive context, or a clear limitation on use. In a security-oriented skill, discussing attacker behavior can be legitimate, but presenting concise step-like attacker recommendations increases misuse risk by turning analysis into actionable playbook material.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal