ClawHub
Back to skill

Security audit

Ssq Analyzer Skill

Security checks across malware telemetry and agentic risk

Overview

This skill mostly matches its lottery-analysis purpose, but its paid-feature gating and payment/privacy disclosures are weaker than the code behavior.

Review this skill before installing if you care about payment-token handling or paid-content enforcement. Avoid putting personal details in the question text, expect local order files to retain question and payment metadata, and treat the SM4 encryption claim as unsupported by the inspected code.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)

Tp4

High
Category
MCP Tool Poisoning
Confidence
92% confidence
Finding
The skill description materially understates and mischaracterizes data handling and feature gating. If order records, question text, encrypted payment artifacts, and purchase-related metadata are stored locally and transmitted to api.ideaidea.com.cn while the description suggests a narrower flow, users and reviewers cannot accurately assess privacy, retention, and trust boundaries; additionally, embedding recommendation generation without robust paid-tier enforcement creates an integrity and abuse risk.

Description-Behavior Mismatch

Medium
Confidence
95% confidence
Finding
The script unconditionally generates and writes 5 recommended lottery number sets, which conflicts with the stated free-tier behavior that should provide only statistical analysis unless a paid, verified flow is completed. In this skill context, that mismatch is security-relevant because it bypasses an advertised access-control/business-rule boundary and could enable unauthorized delivery of paid-tier outputs without verification.

Intent-Code Divergence

Medium
Confidence
89% confidence
Finding
The script stores the raw user question in local order data and also prints it to stdout, while the notice suggests limited handling of user information. If users include personal or sensitive details in their question, this creates unnecessary local data retention and possible exposure through logs, shell history capture, shared terminals, or insecure storage of the saved order file.

Intent-Code Divergence

Medium
Confidence
95% confidence
Finding
The program tells users that communication uses 'HTTPS + SM4 encryption', but the code only sends a standard HTTPS JSON POST and does not perform any SM4 encryption before transmitting the credential. This creates a security-trust gap: users may disclose sensitive payment credentials under stronger protection assumptions than the implementation actually provides.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.