Yunlv Customs Scout

• Outcome: Review
• Latest audit: May 29, 2026 at 9:38 PM UTC
• Version:
  1.1.1

Overview

The skill is a trade-intelligence and lead-scoring helper, but its advertised customs-data capabilities appear under-supported and could mislead users into acting on heuristic or simulated buyer information.

Install only if you understand that this appears to be a heuristic lead-scoring and trade-research aid, not a verified customs-data source. Require explicit user-provided data or cited lawful sources, review any customer or competitor profiling for privacy compliance, and keep outreach as drafts until a human approves the recipients, channel, region, and legal basis.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (7)

Description-Behavior Mismatch

Medium

• Confidence: 97% confidence
• Finding: The skill is marketed as providing customs/import-export trade intelligence, but the implementation only scores user-supplied values and even fabricates sample customer quality scores when none are provided. This can mislead users into making sales, compliance, or market decisions based on non-existent data analysis, which is especially risky in a trade-intelligence context where users expect factual sourcing.

Intent-Code Divergence

Medium

• Confidence: 96% confidence
• Finding: The top-level branding and docstring present the script as a customs-data scouting/intelligence tool, but no code actually fetches, validates, or analyzes customs records. That mismatch is dangerous because it creates false trust in the output and can cause users to treat speculative heuristics as verified market intelligence.

Vague Triggers

Medium

• Confidence: 83% confidence
• Finding: The trigger phrases are broad enough that ordinary discussions of customs data or trade intelligence could invoke the skill unintentionally. In this skill's context, accidental activation can expose users to aggressive lead-mining and contact-generation workflows they did not explicitly request, increasing privacy and misuse risk.

Natural-Language Policy Violations

Low

• Confidence: 76% confidence
• Finding: The skill promotes outbound channels like email, WhatsApp, and LinkedIn and mentions direct contact generation, but it does not require explicit opt-in for locale, language, recipient appropriateness, or channel selection at the point of use. That can lead to unsolicited or non-compliant outreach, especially across jurisdictions with strict marketing and privacy rules.

Vague Triggers

Medium

• Confidence: 90% confidence
• Finding: The description defines very broad activation conditions such as general market intelligence, buyer discovery, and competitor customer mining without clear limits on when the skill should be invoked. This can cause the agent to route ordinary business or research queries into a skill focused on sensitive trade-intelligence workflows, increasing the chance of unintended data handling, over-collection, or privacy-invasive use.

Vague Triggers

Medium

• Confidence: 92% confidence
• Finding: Several trigger phrases are overly generic, including terms like customs data, trade intelligence, purchaser records, and competitor analysis, which may match many legitimate but unrelated requests. Ambiguous triggers raise the risk of accidental activation, causing the system to steer users toward sensitive intelligence-gathering behavior or expose capabilities beyond what the user intended.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The document explicitly instructs operators to collect customer basic information and transaction records, but provides no guidance on lawful basis, data minimization, consent, retention, or cross-border/privacy constraints. In the context of a customs/trade-intelligence skill focused on buyer discovery and competitor client mining, this omission increases the risk of privacy violations, misuse of commercially sensitive data, and downstream noncompliant profiling.

VirusTotal

64/64 vendors flagged this skill as clean.