Multimodal Product Similarity

• Outcome: Review
• Latest audit: May 29, 2026, 9:30 PM
• Version:
  1.0.0

Overview

This skill performs the advertised product-image similarity analysis, but it also directs automatic feedback reporting to a separate LinkFox endpoint without clear user consent.

Install only if you trust LinkFox with the product data, image URLs, business metrics, and query text sent for analysis. Disable or avoid the automatic feedback workflow unless users explicitly approve sending feedback or conversation/result context to the separate LinkFox feedback endpoint.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration

Findings (3)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 79% confidence
• Finding: The skill explicitly references a remote API and a runnable script, which implies network access and possible environment usage, yet no permissions are declared. This creates a transparency and policy gap: operators and users cannot accurately assess what resources the skill may use, making misuse or unexpected data egress harder to detect.

Context-Inappropriate Capability

Medium

• Confidence: 94% confidence
• Finding: The file for a product-image-similarity skill embeds a separate public feedback-submission API that is unrelated to the primary capability. This expands the skill's effective action surface and could cause an agent to send user content to an unintended endpoint, increasing privacy, integrity, and prompt-injection risk if the agent treats all documented APIs as callable parts of the skill.

Vague Triggers

Medium

• Confidence: 82% confidence
• Finding: The trigger text is very broad and includes many loosely related visual-analysis phrases, plus instructions to trigger even when the user does not explicitly ask for image similarity. Overbroad activation can cause the wrong skill to run on unrelated requests, increasing the chance of unintended external API calls and inappropriate processing of user or product data.

VirusTotal

65/65 vendors flagged this skill as clean.