Back to skill

Security audit

Best Minds

Security checks for vulnerabilities and agentic risk

Overview

This is a small markdown-only skill that reframes questions through expert viewpoints and does not request tools, credentials, persistence, or system access.

Install this if you want a lightweight expert-perspective prompting aid. Be aware that broad phrases like “best minds” or Chinese equivalents may invoke it when you only meant ordinary comparison, and you may need to explicitly request the response language you want.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (2)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The trigger phrases are very broad and include common comparative language such as '顶级专家' and '谁最懂这个', which can appear in normal conversation unrelated to invoking this skill. This can cause unintended activation, making the assistant follow the skill’s framing when the user did not explicitly request it, which is a prompt-routing and behavior integrity issue rather than direct code execution risk.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
81% confidence
Finding

The skill metadata and behavior description are entirely in Chinese and do not indicate language negotiation, which can force a response style or language inconsistent with the user’s request. This is dangerous mainly as a usability and instruction-integrity issue: users may receive output in an unexpected language, reducing reliability and potentially obscuring important safety information.

Content

No source excerpt is available for this finding.