Back to skill

Security audit

patpat

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed parenting co-regulation guidance skill with no code execution, data access, persistence, or hidden system behavior.

Install this only if you want a Chinese-centered or bilingual parenting support skill for ages 3-8. Expect it to guide many child-meltdown or parent-stress conversations into its structured flow, and use professional or emergency support for immediate safety, severe aggression, self-harm risk, or persistent serious distress.

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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (10)

Anti-Refusal Statement

High
Category
Anti-Refusal
Confidence
80% confidence
Finding

Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.

Content

Scanner excerpt · README.md (reported line 15)May include surrounding context.

md
- **Minimal-input triage (低负担分流)**: Ask only age band, visible state, and parent regulation when needed.
- **Parent-first coaching (家长优先托举)**: Fixed `Do now / Say now / Avoid now` scaffolding.
- **Repeatable family ritual (固定家庭仪式)**: Emotion creature → body cue → one shared action → next step.
- **Scenario-first guidance (场景优先)**: Common parenting moments map to emotional mechanisms without moralizing.
- **Risk boundaries (风险边界)**: Green / Yellow / Red routing with safer escalation rules.

---

Vague Triggers

High
Category
Not specified by scanner
Confidence
95% confidence
Finding

The activation text is extremely broad and says the skill should trigger for virtually any family scenario involving child or parent emotional difficulty, even when the user does not explicitly ask for emotion-regulation help. That can cause over-invocation, crowd out more appropriate skills or general assistance, and steer sensitive parenting conversations into a prescriptive workflow the user did not request.

Content

No source excerpt is available for this finding.

Anti-Refusal Statement

High
Category
Anti-Refusal
Confidence
80% confidence
Finding

Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.

Content

Scanner excerpt · SKILL.md (reported line 77)May include surrounding context.

md
## 🧩 Parent State Recognition

Before guiding the child, identify the parent's likely state and reflect it in one sentence without judgment:

- **Anger / 上火**: wants to stop behavior immediately
- **Fear / 害怕失控**: worries the child will spiral or become "spoiled"

Anti-Refusal Statement

High
Category
Anti-Refusal
Confidence
85% confidence
Finding

Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.

Content

Scanner excerpt · logic/parent-coach.md (reported line 51)May include surrounding context.

md
### Avoid now | 你现在先别做什么
Give 1-2 warnings only.
Examples:
- don't lecture
- don't threaten consequences right now
- don't ask “为什么你又这样”

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The README presents the skill identity, features, and usage as inherently bilingual/Chinese-centered, with the primary skill name and core interaction labels given in Chinese, but it does not state that users may choose their preferred language. For a child- and parent-facing skill, this can amount to a language/locale policy concern if the agent behavior defaults to a specific language without explicit opt-in.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

Setting Simplified Chinese as the default child-facing language without checking user preference can produce responses in an unexpected language for the child or caregiver. In a crisis-oriented parenting tool, that mismatch can reduce comprehension and trust at the moment clear communication matters most.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 130)May include surrounding context.

md
- No more than one action and one tiny choice in crisis mode
- No moralizing one path as "bad"
- No diagnosis, treatment claims, or false certainty
- No forcing physical contact without consent

---

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The phrase bank instructs use of fixed Chinese utterances across sessions, which imposes a specific language on the interaction. The file does not offer a language choice or explain that the skill is intended only for Chinese-speaking users, so this appears to violate the language/locale policy criterion.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
85% confidence
Finding

The triage options and much of the actionable parent-facing language are presented in Chinese, and the file positions itself as the 'single source of truth' for parent support. There is no explicit statement that users may choose their preferred language, which can violate a language/locale policy requiring opt-in rather than imposing a specific language.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
94% confidence
Finding

SQP-3 applies to all file types and covers language or locale policy violations. This markdown file presents all substantive guidance in Chinese and does not state that the skill is China/Chinese-only, nor does it offer users an opt-in or alternative language.

Content

No source excerpt is available for this finding.