ClawHub
Back to skill

Security audit

Zhang Yidong Macro Analysis

Security checks for vulnerabilities and agentic risk

Overview

This is a plain Markdown investment-analysis skill with no code or tool access, but it gives portfolio-allocation and timing-style guidance despite saying it will not do that.

Review before installing if you expect strict non-advisory financial analysis. The skill has no code execution or data-access behavior, but users may receive actionable-sounding portfolio structure, timing, or allocation language despite the disclaimer that it is not investment advice.

A.I.G

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill states it does not provide buy/sell or position-allocation decisions, but later sections give concrete timing and portfolio-allocation guidance such as '中性仓位', '红利底仓 + 科技弹性哑铃', '可配但需分批', and workflow output fields for仓位/行业/个股. This creates a policy-boundary bypass where downstream agents or users may treat the skill as compliant and then receive actionable investment advice anyway.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The document frames specific references as 'methodological demonstrations,' but repeatedly converts them into operational guidance on when to add exposure, how to structure portfolios, and how to react to market conditions. In context, this is dangerous because the skill is explicitly designed for investment decision support, so these examples are likely to be operationalized rather than read as abstract education.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
90% confidence
Finding

The manifest states "中文为主，英文关键术语保留," which sets a specific language preference. Although the body later allows English on request, the file header itself still encodes a default language constraint that could be seen as a locale policy preference rather than neutral behavior.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
88% confidence
Finding

The Identity section again specifies "中文为主，英文关键术语保留," which reinforces a fixed language policy. While line L021 permits English output if requested, this line by itself still presents the skill as forcing a specific language unless the user opts out.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.