Skill Reviewer Pro

• Outcome: Pass
• Latest audit: May 28, 2026 at 10:16 AM UTC
• Version:
  2.1.1

Overview

This is a disclosed, instruction-only skill for reviewing OpenClaw skills, with no hidden data access, persistence, or destructive behavior found.

Reasonable to install if you want a structured OpenClaw skill reviewer. Before using its validation command, verify the target path is a local skill directory you intended to review, and be aware it may try to mirror the user's detected language even when a clearer explicit language preference would be better.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (3)

Natural-Language Policy Violations

Medium

• Confidence: 95% confidence
• Finding: The skill hard-codes a policy to always respond in the user's language, which overrides explicit user preference and performs behavioral steering without opt-in or justification. While not directly enabling code execution or data exfiltration, it can cause unwanted locale handling and reduce user control over how the agent responds.

Natural-Language Policy Violations

Medium

• Confidence: 96% confidence
• Finding: This section reinforces a mandatory same-language response policy as an unconditional rule, removing operator discretion and user choice. In skill context, this is a policy/control issue rather than a high-severity security flaw, but it is still a true finding because it imposes behavior without consent or justification.

Natural-Language Policy Violations

Medium

• Confidence: 98% confidence
• Finding: The detailed language rules operationalize mandatory language selection and include a fallback based on previous interaction instead of current user choice, which can propagate incorrect assumptions across turns. This is more dangerous than the earlier occurrences because it codifies persistent behavior and consistency requirements that can override present intent and lead to misaligned or privacy-sensitive inference about the user's preferences.

VirusTotal

66/66 vendors flagged this skill as clean.