scl-90

• Outcome: Review
• Latest audit: May 29, 2026 at 9:30 PM UTC
• Version:
  1.0.0

Overview

This appears to be a local offline mental-health self-assessment tool, but it overstates its clinical capability and saves sensitive answers in plaintext without clear opt-in.

Review carefully before installing. This skill does not appear to send data over the network, but do not treat its output as a validated clinical SCL-90 result, and only run it if you are comfortable with plaintext mental-health answers being stored under your home directory.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• 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

Findings (4)

Description-Behavior Mismatch

High

• Confidence: 98% confidence
• Finding: The script claims to provide a professional SCL-90 assessment, but the implemented assessment uses generic numbered prompts rather than the validated questionnaire items and does not compute the advertised factor scores. In a mental-health context, this can mislead users into trusting clinically meaningless output, causing false reassurance or unnecessary alarm and delaying appropriate professional care.

Intent-Code Divergence

High

• Confidence: 97% confidence
• Finding: The help text advertises a complete 90-question assessment and China-norm-based capabilities that the script does not actually deliver. Misrepresentation is especially risky here because users may rely on the tool for sensitive health-related decisions based on incorrect assumptions about completeness and validity.

Context-Inappropriate Capability

High

• Confidence: 99% confidence
• Finding: The script creates a persistent local history directory for storing mental-health assessment results, including raw answers, even though such retention is not necessary for core questionnaire operation. Mental-health responses are highly sensitive personal data, and storing them by default increases the risk of privacy compromise through local account access, backups, shared machines, or endpoint inspection.

Missing User Warnings

High

• Confidence: 99% confidence
• Finding: The script writes detailed assessment results, including the full answer vector, to a history file without warning the user beforehand or obtaining consent. In this context, silent persistence of psychological self-report data is dangerous because users may reasonably expect ephemeral processing and may unknowingly expose intimate health information to others with local access.

VirusTotal

65/65 vendors flagged this skill as clean.