The Girl from the Train

• Outcome: Pass
• Latest audit: Jun 15, 2026 at 10:45 AM UTC
• Version:
  1.0.0

Overview

This is a text-only literary reflection skill with sensitive themes but no executable code, data access, persistence, or hidden install behavior.

Install this if you want a literary lens for discussing trauma, identity, displacement, and rebuilding through this novel. Be aware that it may respond to broad personal disclosures with book-framed emotional guidance, so users seeking clinical, crisis, legal, or professional support should rely on appropriate qualified resources instead.

SkillSpector

By NVIDIA

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 (3)

Vague Triggers

Medium

• Confidence: 94% confidence
• Finding: The trigger condition "Also triggers when the user says they just installed this skill or doesn't know how to start" is overly broad and can cause the skill to activate for users who are not seeking this book-specific framing at all. In a sensitive domain touching trauma, war, identity, and displacement, unsolicited activation risks inappropriate interception of unrelated conversations and ungrounded emotional guidance.

Vague Triggers

Medium

• Confidence: 91% confidence
• Finding: The phrase "I'll show up whenever I sense this book could help" defines activation by subjective model judgment rather than concrete user intent, making invocation boundaries ambiguous. That ambiguity increases the chance of the skill surfacing in unrelated or vulnerable contexts and steering users into a specific narrative lens without a clear request.

Vague Triggers

Medium

• Confidence: 91% confidence
• Finding: The activation scenarios target broad, emotionally common disclosures such as childhood trauma, displacement, loss, and relationship conflict without strong scope boundaries or consent checks. In a sensitive support-oriented skill, this can cause over-triggering on vulnerable users, leading the agent to steer conversations into a predefined therapeutic or worldview-framing response when the user may not have asked for that kind of intervention.

VirusTotal

64/64 vendors flagged this skill as clean.