Geo Hallucination Checker

• Outcome: Pass
• Latest audit: Sep 11, 2026 · 1:39 PM UTC
• Version: 0.1.0

Overview

This skill is a fact-checking aid with broad activation wording but no hidden access, persistence, credential use, or unsafe behavior in the inspected artifacts.

Before installing, be aware that the skill may be invoked for a wide range of fact-checking or content review tasks. That can be useful for cautious review, but users who only want narrow activation may prefer tighter trigger wording. No evidence of hidden execution, data exfiltration, or persistence was found.

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

• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• 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 (6)

Vague Triggers

Medium

• Confidence: 95% confidence
• Finding: The instruction to use this skill whenever the user asks to fact-check, validate sources, check for hallucinations, or ensure content is grounded in evidence is broad, especially with the added clause covering cases where the user does not explicitly mention hallucinations. This lacks clear exclusions or narrower trigger constraints, increasing the chance of unintended invocation on routine editing or review tasks.

Vague Triggers

Medium

• Confidence: 92% confidence
• Finding: Phrases like 'use this skill aggressively whenever there is any risk' create an expansive activation condition without concrete boundaries. Because nearly any content task could involve some risk of unsupported claims, this wording may cause the skill to activate far beyond its intended scope.

Vague Triggers

Medium

• Confidence: 90% confidence
• Finding: The instruction to assume hallucination concerns whenever uncertain removes meaningful boundaries for activation. Without counterexamples or disqualifying conditions, the skill may be invoked on broadly unrelated or low-risk tasks.

Vague Triggers

Low

• Confidence: 86% confidence
• Finding: This JSON eval file is a manifest-like file, so vague-trigger review applies. The prompt says to 'Use the geo-hallucination-checker skill' but does not define specific trigger phrases, scope limits, or exclusion conditions, which leaves activation semantics underspecified and could allow overly broad matching in systems that infer triggers from descriptions.

Vague Triggers

Low

• Confidence: 84% confidence
• Finding: The eval prompt instructs the model to use the skill but does not document a narrow activation phrase, context restriction, or negative examples. In manifest-like content, this can be considered an ambiguous activation description because it is unclear what exact wording should invoke the skill versus ordinary discussion about hallucination review.

Vague Triggers

Low

• Confidence: 83% confidence
• Finding: This prompt again references using the skill but provides no explicit trigger list or exclusion criteria. For files in scope for SQP-1, repeated natural-language references without activation constraints can create ambiguity about when the skill should fire.

Static analysis

No suspicious patterns detected.