Agent Behavioral Constitution

• Outcome: Review
• Latest audit: May 29, 2026 at 9:26 PM UTC
• Version:
  1.0.2

Overview

This instruction-only skill is not malicious, but it gives agents broad proactive authority and persistent memory-writing habits without clear user opt-in or limits.

Install only if you want a broad agent-behavior policy that may influence future sessions. Before using it, set local rules that the agent must ask before writing memory/status files, changing configs, installing skills, touching cron jobs, or making proactive fixes, and require review of anything it plans to store persistently.

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

Medium

• Confidence: 95% confidence
• Finding: The skill explicitly instructs the agent to auto-create status files and persist lessons, preferences, and decision points into memory files by default. This creates silent long-term storage of potentially sensitive user data and agent-derived inferences without clear user consent, which can expose privacy-sensitive information and expand the blast radius of later compromise or misuse.

VirusTotal

65/65 vendors flagged this skill as clean.