ClawHub

Writing Proofreading

Security checks across malware telemetry and agentic risk

Overview

This is a coherent proofreading skill whose optional file edits and Markdown formatting commands are disclosed and aligned with reviewing articles.

Safe to install for article proofreading. Before using automatic formatting or optional external setup, confirm the exact file being changed and review any resulting diff; assess the optional markdown-lint skill separately if you install it.

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
91% confidence
Finding
The skill is presented as a proofreading/review assistant but later instructs operations that can directly modify repository files, including `--fix` formatting commands and initializing another skill. That mismatch can cause an agent to perform write actions in a repo when the user may have expected read-only review behavior, increasing the risk of unintended file changes or over-broad tool use.

Intent-Code Divergence

Medium
Confidence
89% confidence
Finding
The documentation establishes a safety expectation of 'discuss before modifying' but elsewhere includes direct file-changing guidance. Conflicting instructions are dangerous because an agent may follow the more operationally specific command path and bypass the intended confirmation gate, leading to unauthorized or surprising edits.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal