Multi Agent Roles

• Outcome: Pass
• Latest audit: May 28, 2026, 11:53 AM
• Version:
  1.0.0

Overview

This is a descriptive multi-agent role template skill with no executable install steps, credential use, or hidden system access.

Safe to install as a reference/template skill. Before turning its examples into a real multi-agent system, define narrow routing rules, human approval for sensitive domains, and explicit limits on what data each agent may receive or share.

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: 91% confidence
• Finding: The binding routes the strategist agent on any direct message without additional topic or authorization constraints, creating an overly broad activation surface. In a multi-agent system, this can cause unintended invocation, misrouting of sensitive requests, and overexposure of a high-privilege planning role to arbitrary inputs.

Vague Triggers

Medium

• Confidence: 88% confidence
• Finding: The example bindings use generic keywords like content, design, report, test, and bug with no contextual qualifiers, so normal conversation can accidentally trigger specialized agents. This ambiguity can lead to prompt routing errors, unintended data exposure between agent workspaces, and unreliable system behavior when users mention common terms outside the intended workflow.

VirusTotal

66/66 vendors flagged this skill as clean.