ClawHub
Back to skill

Security audit

Avoid AI Writing

Security checks across malware telemetry and agentic risk

Overview

This is a disclosed writing-editing skill that can rewrite or directly edit user-specified text files, with no hidden execution, credential use, or background behavior found.

Install this if you want an agent to audit or rewrite prose for AI-sounding patterns. Use detect mode first for sensitive drafts, and be deliberate with edit mode because it changes the named file in place.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (3)

Vague Triggers

Medium
Confidence
92% confidence
Finding
The README says the skill auto-triggers from phrases like "remove AI-isms," which is broad enough to fire on ordinary editing requests. In an agent environment, overly broad natural-language activation can cause the wrong skill to run, leading to unintended rewrites of user content or tool use when the user did not explicitly request this specialized behavior.

Vague Triggers

Medium
Confidence
95% confidence
Finding
The trigger examples include broad requests like "Clean up AI writing in this paragraph" and "Make this sound less like AI," which overlap with normal editorial instructions. In systems that auto-select skills from natural language, this increases the chance of unintended invocation, misclassification of user intent, and unrequested transformations of text.

Missing User Warnings

Low
Confidence
84% confidence
Finding
The README advertises an edit-in-place mode that modifies files via an Edit tool, but it does not prominently warn that the target file will be changed. In agent workflows, missing warnings around destructive or state-changing actions can lead to accidental modification of user files, especially when paired with broad triggers.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.