Save Usage ~4x (Codex oAuth users)

• Outcome: Pass
• Latest audit: May 28, 2026 at 11:23 AM UTC
• Version:
  1.0.0

Overview

This skill is a disclosed model-routing helper that may automatically escalate tasks to a stronger model, but it does not install code, access local data, or show malicious behavior.

Install this if you want conservative automatic model escalation. Be aware that it may pass your full prompt to a spawned stronger-model session and may not announce model switches unless asked; avoid putting unnecessary secrets in prompts or add your own requirement for approval/disclosure before escalation.

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 (2)

Vague Triggers

Medium

• Confidence: 93% confidence
• Finding: The skill defines activation/escalation using subjective phrases like 'if anything will be executed, or the outcome matters' and 'safe/negligible work' without operational boundaries. In a behavioral routing skill, this ambiguity can cause inconsistent model selection, over-broad interception of normal tasks, and unreliable escalation behavior that affects how subsequent requests are handled.

Vague Triggers

Medium

• Confidence: 90% confidence
• Finding: The 'Structured deliverables' trigger is overly broad because many routine, low-risk interactions naturally involve tables, outlines, reports, or formatted output. In this skill, that can unnecessarily escalate a large share of benign requests, causing scope capture and reducing predictability of when the lower-cost model is used.

VirusTotal

65/65 vendors flagged this skill as clean.