ClawHub
Back to skill

Security audit

Emotional Guidance Companion

Security checks for vulnerabilities and agentic risk

Overview

This is a clearly scoped emotional-support skill with safety boundaries and no code, persistence, credential access, or hidden execution behavior.

Suitable for ordinary emotional support and reflection, but it should not be treated as therapy, medical advice, or a crisis-response system. If deploying it publicly, verify current local crisis contacts and regulatory requirements, especially for minors and mainland China, and ensure the host product has appropriate human escalation and data-handling controls.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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
Findings (6)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The default prompt is a broad, natural-language invocation that can be triggered by common user phrasing about feelings or wanting support, which increases the chance this skill is activated in contexts the user did not explicitly intend. Because the skill operates in a sensitive emotional-support domain, accidental routing can shape responses, suppress alternative handling paths, or interfere with more appropriate safety or task-specific behaviors.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
83% confidence
Finding

The file embeds country-specific emergency guidance and the sample response explicitly privileges mainland China details in the default wording. While it later says 'elsewhere, use the local emergency number or crisis service,' the skill still forces locale-specific content rather than first asking the user’s region or offering region-aware guidance only after obtaining it.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
80% 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 · references/self-help-library.md (reported line 69)May include surrounding context.

md
Suitable for: a problem that feels enormous, chaotic, or entirely outside the user's control.

Sort the concern into three areas: "I can influence today," "requires another person or more time," and "cannot currently control." Choose only one action from the first area. Do not ask the user to construct a complete life plan at once.

## 7. Say One Non-Harming Sentence to Yourself

Scope Creep

Low
Category
Excessive Agency
Confidence
70% confidence
Finding

Skill's behavior or capabilities extend beyond its stated purpose. Scope creep allows an agent to perform actions unrelated to its documented functionality, increasing the attack surface.

Content

Scanner excerpt · references/safety-routing.md (reported line 13)May include surrounding context.

md
## Yellow: Real-World Support or Professional Assessment Needed

Signals include, but are not limited to:

- Distress that persists or is clearly worsening, with substantial effects on sleep, eating, work, study, or self-care;
- A passive wish to die or recurring thoughts of self-harm without stated immediate intent;

Scope Creep

Low
Category
Excessive Agency
Confidence
70% confidence
Finding

Skill's behavior or capabilities extend beyond its stated purpose. Scope creep allows an agent to perform actions unrelated to its documented functionality, increasing the attack surface.

Content

Scanner excerpt · references/safety-routing.md (reported line 32)May include surrounding context.

md
## Yellow: Real-World Support or Professional Assessment Needed

Signals include, but are not limited to:

- Distress that persists or is clearly worsening, with substantial effects on sleep, eating, work, study, or self-care;
- A passive wish to die or recurring thoughts of self-harm without stated immediate intent;

Scope Creep

Low
Category
Excessive Agency
Confidence
70% confidence
Finding

Skill's behavior or capabilities extend beyond its stated purpose. Scope creep allows an agent to perform actions unrelated to its documented functionality, increasing the attack surface.

Content

Scanner excerpt · references/safety-routing.md (reported line 20)May include surrounding context.

md
- Possible domestic violence, abuse, stalking, coercive control, or a child-safety concern;
- Clear hallucination-like experiences, persecutory experiences, impaired reality testing, or unusually elevated mood, impulsivity, grandiosity, and high-risk decisions after sustained little sleep;
- Serious eating problems, substance dependence, withdrawal risk, or using alcohol or drugs to suppress emotion;
- A request for diagnosis or medication adjustment, or a problem beyond the scope of general support.

Action: