Security audit

Game Toolkit Free

Security checks across malware telemetry and agentic risk

Overview

The skill is mostly a game-design prompt pack, but it asks for command execution and file-write capability without clear need or user-controlled scope.

Review this skill before installing. It does not show destructive or exfiltration behavior, but it requests broader authority than its game-design purpose needs. Prefer installing only if write and exec permissions can be disabled or constrained, and only use file export after choosing the destination explicitly.

SkillSpector

By NVIDIA

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration

Findings (5)

Description-Behavior Mismatch

Medium

Confidence: 95% confidence
Finding: The skill presents itself as a local, Markdown-driven game design tool, yet it declares write and exec capabilities that are not necessary for generating game rules. This mismatch expands the attack surface: an agent may permit file modification or command execution under the guise of harmless game generation, enabling unintended side effects if the skill is invoked automatically.

Context-Inappropriate Capability

High

Confidence: 98% confidence
Finding: Granting command execution to a personal game-design skill is unjustified by the documented functionality and creates a powerful primitive that could run shell commands unrelated to the user’s request. In an agent environment, this can lead to arbitrary local command execution, data exposure, or environmental tampering if the skill is misused or prompt-injected.

Intent-Code Divergence

Medium

Confidence: 87% confidence
Finding: The documentation claims the skill works offline with no network requirement, but the error-handling section instructs users or agents to run network diagnostics such as `ping` and references connectivity-dependent behavior. This inconsistency can mislead trust decisions and may trigger unnecessary command execution or network probing in contexts that expected a non-networked skill.

Vague Triggers

High

Confidence: 97% confidence
Finding: The trigger conditions mention data analysis, reporting, statistics, and visualization, which are unrelated to a game-design skill and are broad enough to cause accidental invocation in many unrelated contexts. Overbroad or mismatched triggers are dangerous because they can route arbitrary user tasks into a skill that has exec/write capabilities, increasing the chance of unintended tool use.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The skill describes export/save/convert operations and also declares write capability, but it does not warn users that files may be created or modified. In an agent setting, undocumented write behavior can lead to silent file creation, overwrites, or persistence of generated content without meaningful user consent.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.