ClawHub
Back to skill

Security audit

3D模型审计技能

Security checks for vulnerabilities and agentic risk

Overview

This is a 3D model auditing guide with local helper scripts, but its automated checks are more limited than its broad audit language suggests.

Install only if you want an agent-guided 3D model review workflow. Treat its results as advisory, verify 3MF geometry and legal/content decisions independently for important use, and do not run publish.sh unless you intentionally want to publish this skill package.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)

Description-Behavior Mismatch

Medium
Confidence
98% confidence
Finding
The skill metadata and module comments claim support for 3MF auditing, but the implementation explicitly rejects 3MF files. This creates a security/compliance gap because users may rely on the tool to validate 3MF models and incorrectly treat unvalidated files as covered by the audit process.

Description-Behavior Mismatch

High
Confidence
99% confidence
Finding
The skill promises copyright/IP detection, sensitive-content detection, and printer build-volume checks, but the code only performs basic file-format parsing and geometry extraction. In this context, the mismatch is dangerous because users may depend on the tool for moderation and print-safety decisions that it does not actually perform, leading to policy bypass and unsafe approval of prohibited models.

Intent-Code Divergence

Medium
Confidence
97% confidence
Finding
The docstring states that STL, OBJ, and 3MF are supported, but the code does not support 3MF. Misleading capability claims are especially risky in an auditing skill because operators may assume security checks were performed on unsupported files and make trust decisions based on false assurances.

Missing User Warnings

Medium
Confidence
85% confidence
Finding
The script performs an external publishing action immediately after printing metadata, with no confirmation prompt, dry-run mode, or guardrail. A user running the script by mistake, or with incorrectly parsed metadata, could unintentionally publish or overwrite a marketplace artifact, causing unauthorized release or operational impact.

Static analysis

No suspicious patterns detected.