Morgana Anti Infinite Loop V2 En

• Outcome: Pass
• Latest audit: Jun 9, 2026 at 2:14 AM UTC
• Version:
  2.0.1

Overview

This is a coherent anti-loop helper that runs locally and stores loop fingerprints, but users should be aware it can retain small samples of agent actions on disk.

Install only if local cross-session loop memory is acceptable for your use case. Avoid passing secrets or sensitive customer data in action strings, consider providing a project-specific storage_path, and periodically review or delete ~/.anti_loop/loops.json if you do not want retained agent traces.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (4)

Description-Behavior Mismatch

Medium

• Confidence: 90% confidence
• Finding: The component persists loop fingerprints and action samples to ~/.anti_loop/loops.json across sessions, which creates undeclared local state and may retain sensitive prompt or task content beyond the current run. In an agent context, actions can contain secrets, proprietary data, or user prompts, so cross-session storage expands the data exposure surface beyond what a lightweight runtime guard implies.

Context-Inappropriate Capability

Medium

• Confidence: 86% confidence
• Finding: The code creates directories and writes JSON state to the user's home directory without explicit consent or clear necessity for basic loop detection. Persistent filesystem writes can violate least surprise, leak task history to other local processes/users, and create compliance or privacy issues in shared or managed environments.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The module writes persistent loop data locally without any user-facing disclosure in this file, which means users may unknowingly retain operational traces on disk. Because agent actions often embed user content, prompts, or internal workflow details, silent persistence increases privacy and data-governance risk.

Missing User Warnings

Medium

• Confidence: 84% confidence
• Finding: The file description indicates cross-session loop fingerprints are stored in a persistent local file under the user's home directory. Even if hashed, these fingerprints can create an undeclared behavioral trail across sessions, potentially exposing sensitive usage patterns, enabling correlation of prompts/workflows, or violating user expectations in privacy-sensitive environments.

VirusTotal

65/65 vendors flagged this skill as clean.