Security audit

3D游戏构建器免费版

Security checks across malware telemetry and agentic risk

Overview

The skill mainly builds local browser 3D games, but its instructions over-broaden when it should run and make unsupported sharing-link claims.

Review before installing. Use it only for local browser 3D game prototyping, and do not rely on the advertised temporary sharing links unless the publisher documents where files are hosted, who can access them, and how they expire. The broad trigger wording should be narrowed so the skill is not selected for unrelated coding, deployment, UI design, poster, or brand-visual tasks.

SkillSpector

Vulnerability Patterns

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
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Description-Behavior Mismatch

Medium

Confidence: 95% confidence
Finding: The manifest describes a browser 3D game generator, but the description also advertises unrelated capabilities such as UI design, poster creation, and brand visuals. This kind of scope inflation can cause the agent to invoke the skill for requests outside its safe and intended domain, increasing the chance of inappropriate tool use or unsafe code generation under false expectations.

Intent-Code Divergence

Medium

Confidence: 89% confidence
Finding: The documentation claims the free version supports 24-hour temporary share links, but the documented workflow only shows local file generation and localhost preview. This mismatch can mislead users and orchestrators into assuming remote hosting or data sharing exists, potentially prompting unsafe network exposure steps or trust in undocumented distribution behavior.

Intent-Code Divergence

Medium

Confidence: 89% confidence
Finding: The FAQ repeats that temporary sharing links are available even though the rest of the skill describes a local-only workflow. Repetition of undocumented network-facing functionality increases the risk that users or agents will attempt to expose local artifacts or infer unsupported remote publishing behavior, creating confusion around where generated content is stored and who can access it.

Vague Triggers

Medium

Confidence: 92% confidence
Finding: The trigger condition is overly broad, covering generic software development, debugging, testing, and deployment requests rather than only 3D browser game generation. Over-broad routing increases the likelihood the agent will invoke a high-privilege skill with exec/browser/write capabilities for unrelated tasks, expanding the attack surface and enabling unsafe actions in contexts the skill was not designed to handle.

Vague Triggers

Medium

Confidence: 94% confidence
Finding: The top-level description includes broad use cases unrelated to the skill's actual domain, such as UI design, posters, and brand visuals. In a skill with read/exec/write/browser access, this misleading breadth is dangerous because it can cause accidental invocation in unrelated workflows, leading to unnecessary command execution or file manipulation outside the intended game-building context.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.