Principal Agent Audit

• Outcome: Pass
• Latest audit: Jun 2, 2026, 5:59 PM
• Version:
  0.1.2

Overview

This is a local agent-audit helper that may review sensitive agent context, but its access is disclosed, bounded, and aligned with its purpose.

Install this only if you want a local audit helper that can inspect agent memory, notes, learnings, installed skills, and scheduled-work context. Because implicit invocation is enabled, use clear prompts when requesting audits and require explicit approval before any edits, scheduler changes, authentication changes, publishing, or network lookups.

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 (1)

Vague Triggers

Medium

• Confidence: 89% confidence
• Finding: The default prompt is broad enough to activate on common audit-related phrasing and it authorizes review of sensitive areas like memory, learnings, skills, and scheduled work without explicit scope or approval boundaries. Combined with allow_implicit_invocation: true, this increases the chance of unintended triggering and overbroad inspection of internal state or operational data.

VirusTotal

64/64 vendors flagged this skill as clean.