ClawHub
Back to skill

Security audit

cloud-palace-minimax-h3

Security checks for vulnerabilities and agentic risk

Overview

This is a pure prompt-writing skill for fantasy image and video prompts, with no code execution, data access, persistence, or credential handling.

Installers should expect this skill to automatically format cloud-palace or xianxia-themed image and video requests into English prompt text. In environments with many creative skills, consider narrowing router triggers so generic requests like making a video or poster do not select this skill accidentally.

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
  • 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 (3)

Vague Triggers

Medium
Confidence
91% confidence
Finding
The video-trigger phrases include generic requests such as '做成视频', '动起来', and '生成短片', which are common across many unrelated workflows. This can lead the skill to activate too broadly and steer an agent into producing MiniMax H3-specific prompt output when the user intended a different video tool or a different domain entirely, creating prompt-hijacking and misrouting risk at the orchestration layer.

Vague Triggers

Medium
Confidence
92% confidence
Finding
The image-trigger keywords are broad enough to match many ordinary creative-image requests, which can cause this skill to auto-activate outside its intended niche. In an agent system with multiple skills, that increases the chance of prompt routing errors, unintended instruction takeover, or suppression of a more appropriate skill, even though this file itself does not execute code or access external systems.

Natural-Language Policy Violations

Medium
Confidence
91% confidence
Finding
The instructions state that all rewritten video-prompt paragraphs must be in English, which imposes a specific language choice regardless of the user's preferred language. This is a natural-language policy concern because the file does not offer a language option or document a user opt-in for this constraint.

Static analysis

No suspicious patterns detected.