OEE Content Humanizer

• Outcome: Pass
• Latest audit: May 28, 2026 at 11:17 AM UTC
• Version:
  1.0.0

Overview

This skill does what it claims: it rewrites supplied text through Anthropic Claude, with privacy considerations but no evidence of hidden, destructive, or deceptive behavior.

Install only if you are comfortable sending the text you provide to Anthropic and paying for usage on the Anthropic key in your environment. Avoid using it for confidential, regulated, customer, or proprietary content unless external LLM processing is approved.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (2)

Vague Triggers

Medium

• Confidence: 90% confidence
• Finding: The trigger phrases are generic editing requests that commonly appear in normal user conversations, so an agent that auto-invokes skills based on phrase matching could activate this skill unintentionally. In this skill, unintended activation is more concerning because execution may send user-provided text to an external LLM service via the configured API key, creating unnecessary data exposure and incorrect tool routing.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: User-supplied text is sent off-box to Anthropic, but the CLI does not give a clear user-facing warning at runtime that their content will leave the local environment. This creates a privacy and data-handling risk, especially if users pass confidential drafts, emails, or proprietary material assuming local-only processing.

VirusTotal

66/66 vendors flagged this skill as clean.