Security audit

Market Research Agent

Security checks for vulnerabilities and agentic risk

Overview

This is a text-only market research skill with no executable code, but it is narrowly biased toward Chinese-language reports about the U.S. market.

Install this only if you want a structured Chinese-language market research workflow primarily for the U.S. market. For non-U.S. markets or other output languages, explicitly tell the agent to override the skill's defaults and verify regional assumptions before relying on the report.

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 (8)

Vague Triggers

Medium

Confidence: 89% confidence
Finding: The trigger list includes broad natural-language phrases such as '帮我分析一下XX市场' and general market/product analysis terms that can easily appear in ordinary conversation. This can cause unintended skill activation, injecting this skill's rigid instructions into unrelated contexts and overriding user intent or other more appropriate skills.

Vague Triggers

Medium

Confidence: 92% confidence
Finding: The activation conditions are expansive and load the skill for many loosely related requests, including branding, user insight, and marketing strategy. In a multi-skill environment, this increases the chance of accidental routing, causing the agent to apply an inappropriate fixed workflow and potentially produce misleading or irrelevant outputs.

Natural-Language Policy Violations

High

Confidence: 95% confidence
Finding: Forcing a fixed language and locale without user consent can materially distort outputs by ignoring the user's requested market, audience, or language. In this skill, that constraint is not cosmetic: it affects data sources, cultural assumptions, and strategic recommendations, making the analysis unreliable or misleading for non-U.S. or non-Chinese-output use cases.

Natural-Language Policy Violations

High

Confidence: 98% confidence
Finding: Locking all analysis to the U.S. market regardless of the user's request is a substantive integrity flaw. It can lead to systematically incorrect competitor sets, regulations, consumer behavior assumptions, seasonality, channels, and recommendations, especially when users ask about other countries or global markets.

Natural-Language Policy Violations

High

Confidence: 98% confidence
Finding: The pre-analysis step hard-codes the target market to the United States before any other adaptation, ensuring downstream modules inherit a potentially false premise. Because the skill's entire framework builds sequentially on earlier assumptions, this bias propagates through all nine modules and amplifies the risk of confident but wrong business advice.

Natural-Language Policy Violations

Medium

Confidence: 93% confidence
Finding: The framework hard-codes Chinese as the default output language without checking the user's preferred language. This can cause the agent to ignore user intent or organizational language requirements, reducing reliability and potentially causing misleading or unusable output, but it is not a direct security compromise.

Natural-Language Policy Violations

Medium

Confidence: 95% confidence
Finding: The framework repeatedly fixes analysis to the U.S. market for purchase timing, channels, and data sources, without requiring user opt-in or validating the target market. In a market-research skill, this can systematically produce incorrect recommendations, inappropriate channel selection, and regionally misleading business decisions if the user is researching another geography.

Natural-Language Policy Violations

Low

Confidence: 91% confidence
Finding: The closing note locks the entire framework to a U.S. market perspective, reinforcing a global bias throughout downstream analysis. In this skill's context, that increases the chance of consistently skewed outputs for non-U.S. requests, though the issue is primarily one of safety and quality rather than direct exploitation.

Static analysis

No suspicious patterns detected.