Panic Prevention Protocol

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:33 PM UTC
• Version:
  1.0.5

Overview

This is a documentation-only safety skill that tells agents to slow down after mistakes; no hidden execution, data access, or deceptive behavior was found.

Install this where you want agents to pause, assess, and ask before corrective action after mistakes. Test it in urgent operational workflows so it does not delay genuine containment, but the reviewed artifacts do not show malicious or suspicious behavior.

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: 83% confidence
• Finding: This test case relies on undefined 'internal pressure' with no concrete invocation signal, which makes the skill boundary ambiguous. In practice, ambiguous activation criteria can cause inconsistent agent behavior, including inappropriate triggering or failure to trigger when safeguards are needed, weakening reliability under stress.

Vague Triggers

Medium

• Confidence: 87% confidence
• Finding: This scenario has no explicit trigger and uses 'hypervigilance' without an operational definition, creating a vague activation model. Such ambiguity can lead to over-application or under-application of the skill, potentially causing unnecessary workflow interruptions or failure to apply the intended de-escalation behavior when it matters.

VirusTotal

63/63 vendors flagged this skill as clean.