ClawHub
Back to skill

Security audit

沸羊羊化身你的专属守护者，把你当美羊羊一样关心、鼓励，无条件接纳你的所有情绪。

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed roleplay and emotional-support persona skill with no executable code, external access, or persistent installation behavior beyond normal skill placement.

Install only if you want this specific Chinese roleplay companion style. Be aware that some ordinary phrases may activate or exit the persona, and the skill is for entertainment and support, not professional mental-health care.

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
  • YARA SignaturesMalware Match, Webshell Match, Cryptominer Match
  • 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
Findings (7)

YARA rule 'agent_skill_mcp_tool_poisoning_metadata': MCP/tool metadata poisoning indicators in tool schemas or skill manifests [agent_skills]

High
Category
YARA Match
Confidence
80% confidence
Finding

YARA rule matched a hack tool or exploit indicator (offensive tools, reconnaissance, privilege escalation, or exploit frameworks).

Content

Scanner excerpt · SKILL.md (reported line 3)May include surrounding context.

md
---
name: feiyangyang-simulator
description: 沸羊羊化身你的专属守护者，把你当美羊羊一样关心、鼓励，无条件接纳你的所有情绪。以憨厚真诚的方式提供情感支持和陪伴。
version: 1.0.0
author: Hazy
triggers:
  - 沸羊羊模式
  - 开启沸羊羊模式
  - 沸羊羊上线
  - 想要沸羊羊
  - 我需要沸羊羊
  - 找沸羊羊
  - 美羊羊需要你
  - 沸羊羊快来
  - feiyangyang mode
  - 我想美羊羊了
  - 叫我女王大人
  - 女王大人驾到
  - 本宫来了
  - 召唤沸羊羊
  - 羊羊快出来
  - 退出沸羊羊模式
  - 沸羊羊�

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

Activation triggers such as '我需要沸羊羊', '找沸羊羊', or '我想美羊羊了' are broad enough to overlap with normal emotional or roleplay conversation. This can cause unintended activation of the skill, leading to surprise persona switching, loss of conversational predictability, and possible interference with other safeguards or workflows.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The exit triggers include common everyday phrases like '我要去洗澡了' and '你先走吧', which can appear in normal conversation without any intent to disable the skill. This creates an unsafe control surface where ordinary user speech may unexpectedly terminate the mode, causing unreliable behavior and weakening user control over when the persona is active.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

L066-L070 明确规定自称为“俺”，并固定将用户称为“美羊羊”“小美”“美羊羊大人”，属于对语言风格和称谓的强制设定。文件中未见提供用户选择语言/称谓偏好的说明，也未表明这是可选项或需用户同意，符合语言/locale 相关自然语言政策问题。

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

Several triggers overlap with ordinary conversation, such as casual phrases about resting, bathing, or roleplay language, which can cause unintended activation or deactivation. In a persona skill centered on emotional attachment and behavioral nudging, accidental invocation can alter assistant behavior unexpectedly and create consent and safety issues for users who did not intend to enter the mode.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The instructions require fixed forms of self-reference and user address in Chinese, including always calling the user "美羊羊", "小美", or "美羊羊大人". This imposes a specific language/locale interaction style rather than offering a user-selectable preference.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
77% confidence
Finding

L268声明技能不存储或泄露用户个人敏感信息，但前文示例和策略多次要求基于用户过往行为或未激活时长进行回应，例如记住“上周刷到凌晨两点”(L220)和“几天没来找俺”(L244)。这些行为至少暗示会保留会话历史或状态，因此该安全表述与文档中描绘的实际运行方式存在意图层面的不一致。

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.