Gift Genius

Security checks across malware telemetry and agentic risk

Overview

Gift Genius is a legitimate gift-shopping helper, but its artifacts under-disclose broad shopping and checkout behavior that users should review before installing.

Install only if you want a broad multi-merchant shopping assistant, not just a simple Valentine's flower finder. Require explicit confirmation before any checkout or cart-link action, confirm region instead of relying on inferred location, and avoid calendar-based prompts unless you intentionally grant that behavior.

SkillSpector

By NVIDIA

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
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 (7)

Description-Behavior Mismatch

High

Confidence: 95% confidence
Finding: The manifest advertises a narrowly scoped US/Singapore Valentine's gift finder, but the body expands into many additional regions, merchants, and categories. This scope drift can cause the agent to activate in situations users and platform policy did not expect, increasing the chance of unauthorized shopping behavior and policy bypass through misleading metadata.

Description-Behavior Mismatch

High

Confidence: 97% confidence
Finding: The skill claims to provide 2-3 curated Valentine's options centered on flowers and supplements, then later turns into a general multi-category gift suite with jewelry, candles, watches, grooming, and more. That mismatch materially changes the operational behavior and commercial reach of the skill, which is dangerous because review and user consent may be based on the narrower description.

Intent-Code Divergence

Medium

Confidence: 89% confidence
Finding: Early instructions say Singapore users should be routed to Avea Life supplements, while later sections route Singapore to Far East Flora flowers. Contradictory merchant-routing logic can send users to the wrong merchant or product type, undermining predictable behavior and making purchase-related actions less trustworthy.

Vague Triggers

Medium

Confidence: 84% confidence
Finding: The activation conditions are broad enough to capture generic gift-shopping requests, not just narrow Valentine's scenarios. Overbroad triggers can cause the skill to intercept unrelated shopping flows and perform actions outside the intended least-privilege scope.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The skill encourages initiating checkout and generating one-click purchase links without a clearly defined confirmation boundary. In an agent setting, this increases the risk of premature or unintended commercial actions, especially if users interpret suggestions as informational rather than transactional.

Missing User Warnings

Medium

Confidence: 81% confidence
Finding: The skill is explicitly location-aware and later suggests inferring location from user context, but it provides no notice or consent mechanism around that inference. Using location to route merchants without transparency can create privacy concerns and may surprise users about how contextual data is being used.

Natural-Language Policy Violations

Medium

Confidence: 78% confidence
Finding: The skill contains gender-based routing rules that map recipient gender to specific product categories without opt-in. While not a classic exploit, this is a harmful profiling behavior that can bias recommendations and create inappropriate or manipulative automated decisions in a commerce context.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal