ClawHub
Back to skill

Security audit

Deep Think

Security checks for vulnerabilities and agentic risk

Overview

This is a prompt-only deep-analysis formatting skill; its main risk is over-activation or Chinese-only formatting, not hidden system access.

Install this if you want a Chinese-oriented structured deep-thinking response format. Be aware it may trigger for broad requests for detailed analysis and may steer outputs toward Chinese unless the user or agent overrides the language explicitly.

A.I.G

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

SkillSpector

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 (3)

Vague Triggers

Medium
Confidence
95% confidence
Finding
The trigger definition is overly broad because it includes a non-exhaustive catch-all ('or similar expressions'), which can cause the skill to activate for loosely related requests. In an agent setting, ambiguous activation can override more appropriate skills or cause unintended behavior, especially since this skill changes response structure and reasoning style globally.

Vague Triggers

Medium
Confidence
90% confidence
Finding
The activation section says to trigger whenever the user requests deeper analysis, but it does not define boundaries or exclusions. This ambiguity can cause over-triggering, making the agent apply the skill in contexts where concise answers, safety-specialized handling, or other domain-specific skills would be more appropriate.

Natural-Language Policy Violations

Medium
Confidence
83% confidence
Finding
The skill's required output templates and examples are entirely in Chinese, which implicitly biases or forces output language without checking the user's preference. This can degrade usability, cause incorrect localization behavior, and interfere with downstream systems expecting responses in the user's current language.