ClawHub
Back to skill

Security audit

World Boundry Download

Security checks across malware telemetry and agentic risk

Overview

The skill appears to be a legitimate boundary-data downloader, but it can automatically fall back to GADM's non-commercial data despite contradictory documentation.

Before installing or using this skill, assume the default source chain may use GADM if geoBoundaries fails. For commercial or public-facing work, pin --source geoboundaries or --source natural_earth, verify the reported source/license in outputs, and choose a cache directory you are comfortable clearing.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Description-Behavior Mismatch

Medium
Confidence
97% confidence
Finding
The document states that GADM is only enabled when the user explicitly requests `--source gadm`, but the later selection flow says GADM is automatically used as a fallback. Because GADM has a non-commercial license restriction, this inconsistency can cause users or integrators to unknowingly consume restricted data, creating legal/compliance risk and misleading downstream attribution behavior.

Intent-Code Divergence

Medium
Confidence
96% confidence
Finding
This is a real documentation integrity issue: the prose and decision flow contradict each other about when GADM is selected. In a skill that downloads third-party data with differing license terms, contradictory guidance can directly lead to unauthorized commercial use or incorrect trust assumptions about data provenance.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.