Arya Model Router

• Outcome: Pass
• Latest audit: May 28, 2026, 6:09 AM
• Version:
  0.1.0

Overview

This is a local model-routing helper with disclosed behavior and no evidence of hidden execution, exfiltration, or destructive actions.

Install if you want a local helper for routing tasks to cheaper or stronger models. Review rules.json model names and thresholds, use router auto off or @cheap/@pro overrides when you want control, and avoid forwarding secrets or highly private context into generated briefs unless you intend that brief to be sent to a sub-agent or model.

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: 88% confidence
• Finding: The phrase 'Router: analiza esto' defines an auto mode trigger but does not specify what inputs, contexts, or permission boundaries apply before the router escalates to stronger models or invokes sub-agents. In an agent environment, ambiguous invocation and escalation criteria can cause unintended routing of sensitive prompts or excessive delegation, increasing both data exposure and execution risk.

Vague Triggers

Low

• Confidence: 89% confidence
• Finding: The command triggers are short natural-language phrases like "router auto on" and "router status", which can plausibly appear in ordinary user text or quoted content. If a downstream agent treats these phrases as control commands without strong scoping or confirmation, normal conversation could accidentally change routing behavior or expose internal state.

Natural-Language Policy Violations

Medium

• Confidence: 84% confidence
• Finding: The routing configuration embeds Spanish and English keyword lists that influence model selection and behavior without any explicit user choice, locale negotiation, or policy disclosure. This can cause users in certain languages or domains to be silently routed differently, leading to inconsistent handling, privacy surprises, or degraded safety and quality for multilingual inputs.

VirusTotal

66/66 vendors flagged this skill as clean.