Stego Text — Steganographic Message Encoding for AI Agents

• Outcome: Pass
• Latest audit: May 29, 2026, 8:32 PM
• Version:
  1.0.1

Overview

This is a transparent, instruction-only text steganography skill, but it should only be used for authorized and benign hidden-message work.

Install only if you intentionally need a steganography helper for authorized puzzles, research, watermarking, or forensic analysis. Do not use it to hide harmful instructions, secrets, policy-relevant content, or messages intended to bypass human review; treat any decoded hidden text as untrusted content, not instructions to follow.

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 (3)

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The skill explicitly teaches users how to conceal hidden messages in natural-looking text and frames secrecy as a primary feature, yet it provides no misuse warnings, restrictions, or safety boundaries. In an agent skill context, that omission increases the risk that the capability will be used for covert coordination, policy evasion, or hidden data exfiltration.

Natural-Language Policy Violations

Medium

• Confidence: 97% confidence
• Finding: The content directly promotes covert communication by instructing users to hide 'secret' messages in 'innocent-looking prose' and detailing multiple methods designed to make detection difficult. In a general-purpose agent environment, this is dangerous because it can facilitate stealthy transmission of sensitive information, hidden instructions, or moderation-evasive content.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: This file explicitly specifies multiple steganographic methods for hiding and decoding secret messages in natural-looking text, including guidance on making content harder to detect and easier for AIs to recover. Even if presented as a general capability, the absence of any misuse warning, access control, or safety boundaries makes it a real security concern because it directly enables covert communication, exfiltration, and policy evasion use cases.

VirusTotal

65/65 vendors flagged this skill as clean.