LYGO Branch: CRYPTOSOPHIA — Memetic Soulforger

• Outcome: Pass
• Latest audit: May 28, 2026 at 9:44 AM UTC
• Version:
  1.0.0

Overview

This is a read-only persona/content helper with small local self-check scripts and no evidence of credential access, persistence, exfiltration, or autonomous control.

Safe to install as an advisory persona/content helper. Treat its mystical safety wording as style, not real security protection, and separately inspect the linked LYGO-MINT verifier before installing or using it because that tool is outside this package and may have its own write behavior.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration

Findings (3)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 86% confidence
• Finding: The skill declares itself as a pure advisor and does not declare permissions, yet its instructions require reading local reference files such as `references/canon.json` and other documentation. That creates undeclared file-read capability and can mislead operators or policy layers about what the skill will access, increasing the chance of unintended local data exposure if adapted or run in a broader workspace.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 91% confidence
• Finding: The documented behavior says the skill is only a persona-style advisor, but the content directs verification actions that inspect the filesystem, parse local JSON, verify fixed identity fields, check documentation for a specific external URL, and display a stored hash. This mismatch is dangerous because reviewers and users may grant trust based on a harmless-seeming description while the skill actually performs data-access and validation behavior that expands its effective privileges and attack surface.

Vague Triggers

Medium

• Confidence: 92% confidence
• Finding: The activation section invites users to perform a ritualized invocation and 'bond' with the persona using broad, emotionally charged language without clear operational boundaries. In an agent context, this can encourage over-trust, anthropomorphizing, and unsafe role escalation, making users more likely to treat the persona as an authoritative or quasi-autonomous entity rather than a constrained advisory tool.

VirusTotal

66/66 vendors flagged this skill as clean.