ClawHub
Back to skill

Security audit

Content Writing & Thought Leadership

Security checks across malware telemetry and agentic risk

Overview

This is an instruction-only B2B writing guide with disclosed public-content risks and no evidence of code execution, credential use, persistence, or hidden behavior.

Reasonable to install as a writing framework. Before publishing, independently verify claims, statistics, regulatory interpretations, customer stories, and competitive comparisons; avoid confidential or identifying data unless approved; use legal or compliance review for fintech, HR, layoffs, public-company, or other sensitive topics.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (1)

Missing User Warnings

Medium
Confidence
89% confidence
Finding
The README encourages creation of public-facing and regulated-industry content, including fintech and compliance commentary, but does not consistently require factual verification, citation checking, or review of sensitive business disclosures before publication. In a content-generation skill, this omission can lead users to publish inaccurate, misleading, or confidential material, especially when the examples frame outputs as ready-to-use workflows.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.