Skill Auto Evolver

• Outcome: Pass
• Latest audit: Jun 9, 2026 at 7:37 PM UTC
• Version:
  0.1.2

Overview

This is a local skill monitoring and reporting utility whose file scanning and local logging behavior matches its stated purpose, with privacy cautions for logged context and feedback.

Reasonable to install for local OpenClaw skill monitoring. Before using it, decide which skill directories it may inspect, avoid logging secrets or personal data in context, errors, feedback comments, or user IDs, review exported reports before sharing them, and remember that cleanup deletes old local usage and feedback history.

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)

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The skill explicitly collects usage logs, response times, success rates, and user feedback, yet the documentation does not warn about privacy implications, sensitive metadata capture, retention periods, or operator responsibilities. This can lead to collection of personal or behavioral data without informed consent or appropriate handling, especially when comments or context fields contain sensitive information.

Missing User Warnings

Medium

• Confidence: 87% confidence
• Finding: The documented clear command performs historical data cleanup, but the documentation does not clearly warn that records will be deleted. Users may invoke it assuming it is a harmless maintenance action and unintentionally destroy audit, monitoring, or feedback history needed for troubleshooting or compliance.

Missing User Warnings

Medium

• Confidence: 87% confidence
• Finding: The monitor persists arbitrary `context` and `error_message` data to the database without any minimization, redaction, or consent boundary. In an agent skill, these fields can easily contain prompts, user inputs, secrets, tokens, file paths, or system state, creating a privacy and sensitive-data retention risk if logs are later queried, leaked, or broadly accessible.

VirusTotal

65/65 vendors flagged this skill as clean.