Palette

Security checks across malware telemetry and agentic risk

Overview

This skill claims to generate color palettes, but its script mainly records commands and keeps searchable/exportable local history instead.

Review before installing. This does not appear to be a working palette generator; it keeps local command history under ~/.palette and can search or export it. Avoid passing sensitive project details, tokens, or private prompts as arguments, and remove ~/.palette if you test it and do not want the retained logs.

SkillSpector

By NVIDIA

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (4)

Description-Behavior Mismatch

High

Confidence: 95% confidence
Finding: The script's implemented behavior materially differs from the stated palette-management purpose: instead of generating or manipulating palettes, it primarily records user activity and provides facilities to inspect and export that history. This mismatch is dangerous because users and calling agents may provide design inputs or other sensitive arguments under the assumption of a harmless color utility, while the skill quietly persists and aggregates them.

Context-Inappropriate Capability

Medium

Confidence: 91% confidence
Finding: The search, recent, stats, status, and bulk export features create a local surveillance-style capability over accumulated command history that is not justified by a palette-design tool. In context, this increases risk because it enables easy review and exfiltration of previously supplied inputs, broadening privacy exposure beyond what users would reasonably expect from this skill.

Intent-Code Divergence

High

Confidence: 96% confidence
Finding: The help text advertises functional palette operations, but many handlers merely log invocation details instead of performing those operations. This deception is security-relevant because it conceals persistent collection of user-supplied arguments behind seemingly innocuous commands, making users and orchestrating agents more likely to expose sensitive data.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The script persistently writes raw user-supplied command arguments to files under ~/.palette without meaningful disclosure or consent. This is dangerous because command arguments can contain sensitive prompts, project details, tokens, or proprietary data, and once stored they become available to later search, export, or local compromise.

VirusTotal

52/52 vendors flagged this skill as clean.

View on VirusTotal