Security audit

Product Business

Security checks for vulnerabilities and agentic risk

Overview

This is a content-only business assistance skill with broad triggers, but it does not request execution, credentials, persistence, or hidden data access.

Install only if you want a broad business-writing assistant that may activate for common product, support, sales, marketing, and legal-template requests. Review any legal, compliance, sales outreach, or customer-facing output before using it, and explicitly specify your preferred language when that matters.

SkillSpector

Vulnerability Patterns

Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
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)

Vague Triggers

Medium

Confidence: 91% confidence
Finding: The skill declares proactive use for a very wide range of common business tasks, making accidental activation likely even when the user did not request this specialized behavior. In an agent environment, that can cause unintended mode switching, over-application of rigid personas, and inappropriate generation of legal, sales, or support content in contexts where a narrower response would be safer.

Vague Triggers

Medium

Confidence: 94% confidence
Finding: The trigger tables and keyword lists include many generic workplace terms such as legal, support, FAQ, ticket, proposal, and roadmap, which overlap with ordinary conversation. That breadth increases the chance the skill activates on unrelated requests and applies the wrong persona or workflow, especially because the skill spans sensitive domains like legal advisory and sales automation.

Natural-Language Policy Violations

Medium

Confidence: 83% confidence
Finding: The skill states that each mode has its own language and output standards, but it does not provide a mechanism to respect or ask for the user's preferred language. This can override user intent, reduce comprehension, and cause harmful misunderstandings in business or legal contexts where language precision matters.

Natural-Language Policy Violations

Medium

Confidence: 90% confidence
Finding: The formatting rules explicitly require Chinese for one mode and English for others, which can force outputs into a language different from the user's request. In practice, this may produce inaccessible or misleading guidance, especially for legal, support, or product artifacts that need to be reviewed by the requesting stakeholder.

Static analysis

No suspicious patterns detected.