Sub-agent Task Queue

• Outcome: Review
• Latest audit: May 29, 2026 at 9:28 PM UTC
• Version:
  1.0.0

Overview

The skill is transparent about managing concurrent sub-agents, but it encourages broad automatic spawning of parallel work with only limited guardrails.

Install only if you want your agent to run multiple sub-agents concurrently. Lower the concurrency defaults, require confirmation before starting new sub-agents unless explicitly requested, and store the queue file in a private user-owned location if task names may reveal sensitive work.

SkillSpector

By NVIDIA

Vulnerability Patterns

• 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
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (1)

Missing User Warnings

Low

• Confidence: 93% confidence
• Finding: The setup uses shell redirection to write directly to a predictable world-accessible path in /tmp, which can overwrite an existing file and is vulnerable to symlink or race-condition abuse on multi-user systems. In this skill's context, the file stores only queue metadata rather than secrets or code, so the primary risk is integrity/disruption rather than severe compromise, but it is still an unsafe operational pattern.

VirusTotal

62/62 vendors flagged this skill as clean.