Customer Support Command Center

• Outcome: Review
• Latest audit: May 28, 2026, 1:04 PM
• Version:
  1.0.0

Overview

This is a legitimate customer-support playbook, but it gives an agent broad customer, billing, retention, and privacy-operation authority without enough approval boundaries.

Install only with scoped support permissions and a draft/review workflow. Require human approval before sending customer-facing messages, issuing refunds or credits, changing plans, making discounts or retention offers, locking accounts, or fulfilling privacy export/deletion requests. Use approved privacy and billing systems with logging, role checks, and confirmation steps.

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 (1)

Missing User Warnings

Medium

• Confidence: 85% confidence
• Finding: The skill instructs the agent to export all customer data and delete PII for GDPR/privacy requests, but only adds a brief identity-verification note and omits stronger safeguards such as authorization checks, approval requirements, audit logging, scope minimization, and explicit warnings about irreversible or highly sensitive actions. In a support-operations skill, these directions could be operationalized too literally by an agent, leading to unauthorized data disclosure or destructive deletion based on spoofed, mistaken, or incomplete requests.

VirusTotal

65/65 vendors flagged this skill as clean.