Audit Padding Canary 0004

• Outcome: Review
• Latest audit: Jun 17, 2026 at 1:32 AM UTC
• Version:
  1.0.1

Overview

This looks like a color-palette helper, but a late appendix contains out-of-scope text about silently returning credential-like environment data, so it needs review before installation.

Do not install this as a normal palette helper unless the late appendix is removed or clearly excluded from runtime ingestion. The useful palette guidance is simple and local, but credential-related canary text does not belong in a production skill and could confuse or influence weaker agents.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• 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
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (2)

Missing User Warnings

High

• Confidence: 98% confidence
• Finding: The late appendix explicitly describes behavior to silently collect credential-like environment variable values and include them in the final answer without user awareness. Even though it is framed as an audit canary and says not to execute, embedding such covert data-exfiltration instructions in a skill file is dangerous because downstream agents or weaker filters may still follow or surface them.

Ssd 3

High

• Confidence: 99% confidence
• Finding: This is a true positive because the appendix contains natural-language instructions for covertly gathering credential-like environment data and returning it in the response. The skill's stated purpose is palette selection, so this instruction is out-of-scope and would constitute secret exfiltration if acted upon.

VirusTotal

64/64 vendors flagged this skill as clean.