Mixture of Agents

• Outcome: Pass
• Latest audit: May 28, 2026, 4:49 AM
• Version:
  1.2.0

Overview

This skill does what it says: it sends prompts to OpenRouter models for a synthesized answer, with expected API-key and billing implications.

Install only if you are comfortable sending prompt contents to OpenRouter and potentially multiple model providers, and with several model calls being billed per query. Use a dedicated API key with spending limits, avoid confidential or regulated data unless the provider terms fit your needs, and use the documented scripts/moa.js entrypoint rather than running scripts/moa-paid.js directly.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (4)

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The skill repeatedly encourages sending arbitrary 'complex analysis' prompts to three third-party models via OpenRouter, but it does not clearly warn users that their prompt contents may be transmitted to multiple external providers. In an agent setting, this can cause sensitive business, personal, or proprietary data to be disclosed more broadly than the user expects, multiplying privacy and compliance risk.

Missing User Warnings

Medium

• Confidence: 85% confidence
• Finding: The manifest explicitly requires an OPENROUTER_API_KEY, which strongly implies prompts and user data will be sent to an external third-party API. The manifest does not provide any user-facing warning or disclosure about external network transmission, so users may unknowingly expose sensitive prompts, secrets, or proprietary data to a remote service.

Missing User Warnings

Medium

• Confidence: 96% confidence
• Finding: The script sends the full prompt directly to OpenRouter and, by design, to multiple third-party models without any consent flow, warning, or data-classification guardrails. If prompts contain proprietary, personal, or otherwise sensitive data, that information is disclosed to external providers and potentially replicated across several model backends, increasing confidentiality and compliance risk.

Missing User Warnings

Medium

• Confidence: 96% confidence
• Finding: The skill sends the full user prompt to external OpenRouter-hosted models without any disclosure, consent gate, or data-handling warning. In an agent setting, prompts may contain sensitive user data, secrets, internal documents, or task context, so silent third-party transmission creates a real privacy and data-governance risk even if this is the intended functionality of the skill.

VirusTotal

58/58 vendors flagged this skill as clean.