ClawHub

Data Type Classifier

Security checks across malware telemetry and agentic risk

Overview

This skill appears to be a straightforward construction-data classifier with only ordinary local file/export behavior.

Safe to install for its stated purpose. Use it with construction/project files you intend to analyze, review export locations before writing reports, and note that the included sample saves classification_report.md in the working directory.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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)

Natural-Language Policy Violations

Low
Confidence
78% confidence
Finding
The book reference includes a Russian title first ("Типы данных в строительстве") alongside the English translation, but the document does not state whether the skill expects or prefers Russian-language inputs or outputs, nor does it offer a user language choice. Under the language/locale policy, unexplained language steering in natural-language instructions can be a policy concern when not clearly justified.

Unrestricted Tool Access

Medium
Category
Excessive Agency
Content
- **Format:** {c.detected_format.value}
- **Structure:** {c.structure.value}
- **Storage:** {c.storage_recommendation.value}
- **Tools:** {', '.join(c.processing_tools[:3])}
- **Confidence:** {c.confidence:.0%}
"""
Confidence
85% confidence
Finding
Tools:*

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal