Dianping Query Moss

• Outcome: Review
• Latest audit: May 29, 2026 at 9:24 PM UTC
• Version:
  1.0.0

Overview

This is a restaurant lookup skill, but it asks the agent to use a named Dianping account and potentially handle phone/SMS login while forcing a Hangzhou location workaround.

Review before installing. Use this only if you intentionally want Dianping searches scoped around Hangzhou and are comfortable with the agent using the named Dianping account. Do not paste SMS codes or credentials into chat; complete any login directly in Dianping's own UI and confirm the city/account context before browsing.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (2)

Natural-Language Policy Violations

Medium

• Confidence: 92% confidence
• Finding: The skill hard-codes use of a specific logged-in account ('一定S') and a fixed regional context without user opt-in or an opportunity to choose a different account or locale. This can cause actions and data retrieval to occur under the wrong identity or jurisdictional context, creating privacy, consent, and account-misuse risks.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The skill instructs the agent to use a specific authenticated Dianping account to collect restaurant information, but it does not include a user-facing warning that browsing and queries may occur under that account and may expose account state, location assumptions, or activity history. The additional guidance to 'bypass' location behavior via URL parameters increases concern because it encourages operating around account context rather than transparently obtaining user consent.

VirusTotal

66/66 vendors flagged this skill as clean.