Security audit

Universal Business Skill System

Security checks across malware telemetry and agentic risk

Overview

This is a broad Markdown-only business workflow skill; it can involve sensitive business or personal data, but I found no hidden code, credential use, exfiltration, or privileged system access.

Install only if you want a broad business workflow framework. Use it with authorized, minimum-necessary data; avoid raw personal, customer, regulated, financial, or confidential data unless you have consent or another lawful basis, access controls, and a retention plan. Manually review outputs before using them for CRM updates, public distribution, compliance, financial, customer-facing, or profiling decisions.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (8)

Vague Triggers

High

Confidence: 95% confidence
Finding: The trigger list is extremely broad and consists of common business terms such as '业务', '运营', '内容', '数据', and '流程', which can cause the skill to activate during many unrelated conversations. This creates an invocation-surface problem: users may be routed into this skill unintentionally, causing unwanted instruction loading, context hijacking, or inappropriate workflow execution across a wide range of benign requests.

Vague Triggers

High

Confidence: 94% confidence
Finding: Repeating the same ambiguous trigger catalog in the body reinforces a design that lacks scope constraints, exclusion rules, or examples of when not to invoke the skill. In context, this skill claims to be universally applicable across industries and business flows, so broad triggers become more dangerous because they maximize accidental routing into a very expansive instruction set.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: This section explicitly directs collection of participant information, interaction data, and feedback, but does not require consent, notice, purpose limitation, or handling controls. In an event-operations skill, this can normalize gathering personal data without lawful basis or participant awareness, creating privacy, compliance, and misuse risk.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: The activity data collection workflow calls for collecting, cleaning, and reporting participation, interaction, conversion, and feedback data without any safeguards on personal data handling. Because this is a reusable business skill template, the omission can propagate broad over-collection and unclear retention across many deployments.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: Photo capture and archival can involve identifiable images and metadata, yet the skill only instructs users to photograph, organize, and archive materials without consent, usage restrictions, or retention guidance. This exposes participants to surveillance-like collection, unauthorized reuse, and long-term storage of personal data.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: This section explicitly describes collecting and analyzing user behavior data, including behavioral patterns and paths, but provides no privacy guardrails, lawful-basis checks, data minimization, retention limits, or consent/notice requirements. In a reusable business skill system, that omission can normalize broad behavioral surveillance and lead downstream agents or operators to process personal data in ways that violate privacy expectations or regulation.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: This section goes beyond analytics into user segmentation and profiling, which can enable sensitive inference, unfair targeting, or discriminatory treatment if performed without constraints. Because the skill defines profiling as a standard output without warnings or approval gates, it increases the risk that operators create detailed user profiles from behavioral data without assessing privacy, fairness, or regulatory implications.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The skill explicitly supports collecting customer basic information and maintaining a customer information database, but it provides no privacy safeguards, data minimization rules, consent requirements, retention limits, or handling guidance for potentially sensitive personal or business data. In a customer-operations context, this omission can lead users or downstream agents to store excessive personal data, mishandle regulated information, or create unsafe automated CRM workflows.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.