Security audit

dundun-10agents

Security checks for vulnerabilities and agentic risk

Overview

This is a plain Chinese legal-risk review skill with broad activation wording, but it has no hidden code, data access, persistence, or privileged behavior.

Install this only if you want a Chinese legal-risk review assistant that may activate whenever legal terms appear. Treat its output as drafting and risk-spotting help, not a substitute for a licensed lawyer, and avoid sharing highly sensitive legal documents unless you are comfortable sending them to the model.

SkillSpector

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
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 (2)

Vague Triggers

Medium

Confidence: 92% confidence
Finding: The activation rule is explicitly mandatory and very broad: it says any legal-related matter or legal text must immediately trigger this skill, with a long keyword list that includes common terms like “风险” and “条款”. This can cause inappropriate routing, override more suitable skills, and expose users to unauthorized or unwanted legal-style guidance in contexts that only loosely mention legal concepts.

Natural-Language Policy Violations

Medium

Confidence: 81% confidence
Finding: The skill metadata and content are written to operate in Chinese and provide a fixed Chinese opening/output style without offering a user language preference path. This can reduce usability, create misunderstanding for non-Chinese users, and in legal contexts may increase the chance of misinterpretation of important advice or contract analysis.

Static analysis

No suspicious patterns detected.