胡田 OPC导师 股权合作机制

• Outcome: Review
• Latest audit: May 28, 2026, 11:30 AM
• Version:
  1.0.0

Overview

This is a documentation-only skill, but it needs Review because it gives operational guidance for legal, tax, banking, profit-sharing, and compliance wording workflows without enough safeguards.

Install only if you want a high-level drafting/reference aid for cooperation governance. Do not use it to operate real accounts, move money, solicit participants, publish investment-like materials, store raw contract or financial data on-chain, or share ledgers with other skills without qualified legal/tax/compliance review and explicit participant consent. Ignore or remove the euphemistic sensitive-word substitution guidance; communications should describe the real legal and economic substance plainly.

SkillSpector

By NVIDIA

Vulnerability Patterns

• 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
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (4)

Description-Behavior Mismatch

Medium

• Confidence: 82% confidence
• Finding: The document goes beyond generic collaboration guidance and explicitly frames how to structure participation so it avoids looking like illegal fundraising, including restricting to 'small circles' and 'specific partners.' That is dangerous because it can normalize regulatory-evasion patterns and provide operators with a playbook for packaging fundraising-like activity as private cooperation rather than removing the risky business model.

Context-Inappropriate Capability

Medium

• Confidence: 94% confidence
• Finding: This section is effectively a terminology-substitution guide for investment, securities, fundraising, and MLM-adjacent language, telling users which risky terms to avoid and what safer-sounding alternatives to use. In context, that is dangerous because it can help conceal the true economic nature of an offering from users, auditors, or regulators instead of reducing the underlying legal and fraud risk.

Context-Inappropriate Capability

High

• Confidence: 87% confidence
• Finding: The document prescribes operational controls for handling pooled money and oversight, including tax automation, dual-control bank credentials, broad transaction notifications, daily reconciliation, and immutable evidence storage. For a collaboration skill, these are unusually specific financial-control capabilities that could enable operation of a quasi-financial platform and make risky fund-handling workflows easier to scale and legitimize.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The harness section explicitly instructs the agent to create directories, regenerate missing reference files, patch protocol documents, and reorder content, but the skill does not prominently warn users that it may write or modify local files. In an agent environment, undocumented file creation or mutation can lead to unintended workspace changes, overwriting user content, or silently producing artifacts the user did not consent to.

VirusTotal

64/64 vendors flagged this skill as clean.