ClawHub
Back to skill

Security audit

Payment Integration Guide

Security checks across malware telemetry and agentic risk

Overview

This is a coherent payment-integration guide, but its webhook debugging advice could lead users to log sensitive payment or customer data.

Review the webhook logging guidance before installing or relying on this skill. The integration examples are otherwise aligned with the payment-guide purpose, but do not copy the raw logging pattern into production; log minimal metadata, redact headers/signatures/tokens/customer data, use sandbox credentials first, and require review before enabling live payment flows.

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 (1)

Intent-Code Divergence

Medium
Confidence
96% confidence
Finding
The skill tells users to log raw webhook payload data for debugging after earlier warning not to log sensitive payment data. Even though the sample truncates the body, headers and payloads can still contain personal data, tokens, addresses, or other payment-related metadata that may be exposed through centralized logs, support tools, or log retention systems.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.