LYGO Glyph2Resonance

Security checks across malware telemetry and agentic risk

Overview

This skill performs the advertised local glyph-to-audio workflow, with a disclosed local memory integration that users should understand before using sensitive inputs.

Use this skill for local creative audio generation, but do not run it on private or sensitive glyph images unless you are comfortable with output WAV/JSON files and possible 3-Brain memory entries containing the image name, path, preset, and derived metrics. Review generated profiles before sharing them.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (4)

Description-Behavior Mismatch

Medium

Confidence: 94% confidence
Finding: The skill performs an additional side effect beyond its primary sonification purpose by automatically importing and writing a summary into a separate local memory subsystem ('3-Brain') whenever that module is available. This is dangerous because it silently persists user-derived data without explicit consent, making data flow less predictable and potentially exposing file names, inferred metadata, and workflow information to another component with different retention or access properties.

Intent-Code Divergence

Low

Confidence: 80% confidence
Finding: The documentation states a local-first, review-before-external-use posture, but the code later performs an automatic write into another subsystem without a review gate. Even though the target appears local, the mismatch between stated behavior and actual side effects undermines user trust and can lead operators to run the tool under incorrect assumptions about where outputs are being stored.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: The skill states that runs may automatically grow profiles and metadata into a 3-Brain memory system, but it does not clearly warn that input-derived data may be retained beyond the immediate task. This creates a privacy and data-governance risk, especially if uploaded glyphs, file paths, symbolic content, or generated profiles are sensitive and become searchable or reusable by downstream agent systems.

Missing User Warnings

Low

Confidence: 76% confidence
Finding: The generated profile JSON stores the source image path and extracted analysis metadata by default, which can reveal local directory structure, file naming conventions, or sensitive project context if the profile is shared or ingested elsewhere. In this skill's creative-processing context, metadata export is expected, but the absence of a clear warning or minimization control still creates a privacy and data-leakage risk.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal