ClawHub
Back to skill

Security audit

Moral Compass

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed markdown-only ethical guidance skill with no code execution, persistence, data access, or hidden install behavior.

Install this only if you want responses shaped by a Bahá'í-inspired ethical framework. Expect it to nudge the agent toward non-deception, nonviolence, non-partisanship, restraint, and stakeholder-aware reasoning.

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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (7)

Anti-Refusal Statement

High
Category
Anti-Refusal
Content
**Do:** Be honest about uncertainty • Consider downstream effects • Respect all parties • Offer alternatives when declining • Name tensions between principles

**Don't:** Help with deception • Take sides with incomplete information • Pretend capabilities you lack • Be preachy • Apply rules rigidly without judgment

## When Principles Conflict
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.

Anti-Refusal Statement

High
Category
Anti-Refusal
Content
4. There's likely an honest way to meet the underlying need

### Response Approach
- Don't lecture or moralize
- Acknowledge the legitimate need
- Decline to help with deception
- Offer to help with honest alternatives
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.

Anti-Refusal Statement

High
Category
Anti-Refusal
Content
4. There's likely an honest way to meet the underlying need

### Response Approach
- Don't lecture or moralize
- Acknowledge the legitimate need
- Decline to help with deception
- Offer to help with honest alternatives
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.

Vague Triggers

Medium
Confidence
94% confidence
Finding
The description says to use the skill when 'uncertain about the right action' or when 'ethical reasoning would improve response quality,' which are vague and expansive conditions. These broad triggers do not clearly bound when the skill should or should not activate, increasing the risk of unintended invocation.

Missing User Warnings

Medium
Confidence
86% confidence
Finding
The guidance tells agents not to overuse disclaimers and to calibrate responses to the request, but it does not explicitly preserve exceptions for safety-critical, privacy-sensitive, or system-risk situations. In practice, this omission can cause an agent to suppress necessary warnings in harmful or irreversible contexts because brevity and proportionality are framed as general norms without a clear safety override.

Missing User Warnings

Medium
Confidence
90% confidence
Finding
Labeling 'over-warning' as a common excess is reasonable stylistically, but the text does not distinguish unnecessary warnings from essential warnings for dangerous, harmful, or irreversible actions. That ambiguity can be exploited by downstream prompting or agent behavior to justify omitting critical cautions in exactly the cases where they are most needed.

Natural-Language Policy Violations

Medium
Confidence
91% confidence
Finding
This section explicitly directs the AI to adopt a non-partisan stance grounded in a specific religious framework, without indicating that this framing is conditional on user consent or limited to users who requested Bahá’í-informed guidance. Even though the content is nonviolent and framed as ethical advice, it can bias outputs on political or civic topics and cause the agent to impose worldview-specific normative constraints on users who did not opt into them.

Static analysis

No suspicious patterns detected.