A Share Portfolio Optimize

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:32 PM UTC
• Version:
  1.0.0

Overview

This is a local A-share portfolio analysis skill with some financial-safety documentation gaps, but no evidence of hidden access, trading, persistence, or data exfiltration.

Install only if you are comfortable using it as informational portfolio-analysis tooling, not personalized investment advice. Verify the separate cn-stock-data helper and Python dependencies, be cautious with Black-Litterman claims unless the agent explicitly handles the needed view/confidence inputs, and include a risk disclaimer even in brief outputs.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• 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 (3)

Description-Behavior Mismatch

Medium

• Confidence: 93% confidence
• Finding: The skill advertises Black-Litterman support in its manifest and workflow, but the documented command interface only exposes min_var, max_sharpe, risk_parity, and equal_weight, with no BL-specific parameters for views or confidence. This mismatch can cause the agent to claim unsupported analysis, silently substitute another method, or produce fabricated BL-style output, which is especially risky in a financial decision-support context where users may rely on the stated methodology.

Vague Triggers

Medium

• Confidence: 82% confidence
• Finding: The trigger list includes broad natural-language phrases such as '怎么分配权重' and generic optimization terms that may appear in ordinary financial discussion, increasing the chance of unintended activation. In an investment skill, accidental invocation can steer a conversation into authoritative-looking portfolio advice without the user explicitly requesting quantitative optimization, creating confusion and possible overreliance.

Natural-Language Policy Violations

Low

• Confidence: 88% confidence
• Finding: The skill requires a disclaimer only for the formal style and explicitly omits it for the brief style, creating inconsistent safety messaging based solely on output format. Because this skill produces investment allocation recommendations, users receiving the brief mode may not be warned that results are model-based, historical-data-driven, and not predictive, increasing the chance of misplaced trust.

VirusTotal

66/66 vendors flagged this skill as clean.