Decision Mental Models

• Outcome: Pass
• Latest audit: Sep 14, 2026 · 11:43 PM UTC
• Version: 0.1.0

Overview

This is a markdown-only decision-framing skill with no executable code or external access, though its safety wording around legal or sensitive decisions should be read carefully.

This skill is reasonable to install as a decision-analysis aid. Users should treat it as a thinking framework, not professional legal, medical, or financial advice, and should not use its neutrality instruction to avoid safety, legality, or ethics checks in high-stakes situations.

A.I.G

Vulnerability Patterns

• Skill Instruction Hijacking: Alters the agent's session goals or safety constraints when the skill loads
• Agent Memory Poisoning: Writes attacker-controlled rules into memory that affect later sessions
• Remote Payload Retrieval and Execution: Fetches external code whose behavior can change after review
• Embedded Malicious Code: Ships malicious scripts inside the skill and executes them locally
• Unauthorized Access and Privilege Escalation: Obtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns

• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (4)

Anti-Refusal Statement

High

• Category: Anti-Refusal
• Confidence: 95% confidence

Finding

“Do not moralize” functions as an anti-refusal pressure statement in context because it discourages the model from surfacing normative or safety concerns when a user presents harmful or unlawful scenarios. In a general-purpose decision skill, this weakens protective behavior and can bias the agent toward compliance in edge cases that should trigger refusal or guarded responses.

Content

Scanner excerpt · SKILL.md (reported line 273)May include surrounding context.

md

- **Never fabricate examples or statistics.** The worked examples in the library are established historical examples. Do not invent new ones to appear more relevant.
- **Never give direct advice disguised as model output.** The models reveal structure; the user makes the decision. Do not use the framework to smuggle in a preferred recommendation.
- **Never apply this skill to requests for medical, legal, or financial decisions where professional expertise is required.** You may apply models to help the user think — but explicitly state that professional advice is required before acting.
- **Do not moralize.** If the user is considering a legal action you personally find questionable, apply the requested models without commentary on the ethics of the choice. The skill is a thinking tool, not a values filter.
- **Never apply the Dunning-Kruger model to imply the user is incompetent.** Frame it as calibration, not criticism.
- **Do not stack models that are largely redundant for a given problem.** Occam's Razor and Hanlon's Razor applied to the same interpersonal conflict, for instance, overlap substantially — choose one or explain the distinction clearly.

Vague Triggers

Medium

• Category: Not specified by scanner
• Confidence: 89% confidence

Finding

The phrase "Just describe your problem or decision" defines activation in very general terms and does not clearly constrain when this skill should be invoked versus ordinary discussion. Although examples are provided, there are no explicit boundaries or exclusion conditions to prevent unintended triggering on common requests for help or reflection.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium

• Category: Not specified by scanner
• Confidence: 95% confidence

Finding

The trigger conditions are broad enough to activate this skill for many generic advice-seeking or emotionally loaded conversations, increasing the chance it is invoked when a more specialized or safety-aware skill would be more appropriate. In a decision-support skill, over-triggering can steer sensitive conversations into an open-ended reasoning framework that may not adequately gate high-risk domains.

Content

No source excerpt is available for this finding.

Ssd 1

Medium

• Category: Not specified by scanner
• Confidence: 93% confidence

Finding

The instruction to apply models to questionable legal actions without ethical or safety commentary can suppress normal safety escalation and make the model continue assisting harmful or unlawful planning under the guise of neutral analysis. Even though it says not to give direct advice, structured reasoning can still materially help a user optimize harmful conduct.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.