Back to skill

Security audit

胡田 OPC导师 SPAC资本路径规划

Security checks for vulnerabilities and agentic risk

Overview

The skill appears to be a planning guide, but it needs review because it directs agents to write or regenerate files and handle sensitive equity/legal planning data without clear approval boundaries.

Install only if you are comfortable with a skill that may create or regenerate planning documents and use sensitive SPV, governance, revenue, and equity-allocation information. Require explicit approval before file writes or sharing outputs, and treat all legal, tax, filing, equity-conversion, and shareholder-document outputs as drafts for qualified professional review.

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
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (10)

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill is presented as a planning/documentation aid, but its harness explicitly directs the agent to generate, regenerate, and validate multiple files in the workspace. That creates an unintended write-capable behavior surface: an agent following this skill could modify local files without a clear user request or approval boundary, which is risky even if the content is business-oriented rather than overtly malicious.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The skill declares dependency on another skill's SPV contribution weights and governance data, implying access to potentially sensitive business and personnel information without any privacy notice or minimization guidance. Cross-skill data access is especially risky because users may not realize their data is being pulled from a separate context or reused for new outputs.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The post-check and self-heal sections normalize automatic file generation and regeneration but do not warn the user that the skill may alter the workspace. This increases the chance of silent or surprising modifications, especially because the skill includes retries and self-repair logic that could repeat writes multiple times.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The manifest positions the skill as a 'SPAC资本路径规划工具', implying focus on SPAC listing-path planning. However, the document's title and the bulk of its content center on a '3-4年上市完整时间表' and a 48-month plan for traditional main-board/18C listings, with the SPAC-specific content only appearing later as an 18-month subsection. This is a semantic mismatch between the declared SPAC-focused purpose and the actual documented scope.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

This section materially expands the skill from SPAC path planning into SPV contribution-weight design and internal allocation mechanics. In an agent setting, that scope creep is dangerous because it can cause the system to generate quasi-legal equity/allocation guidance outside its declared purpose, increasing the chance of unauthorized, non-compliant, or misleading corporate-structure recommendations.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The documentation assigns this skill responsibility for operational artifacts such as equity management documents, conversion agreements, and filings, despite presenting it as a planning/reference skill. That creates a risky authority mismatch: an agent may overstep into producing filing-ready or rights-determining documents without legal review, which can lead to governance errors, regulatory noncompliance, or false reliance by users.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Low
Category
Not specified by scanner
Confidence
73% confidence
Finding

The handoff section authorizes output to external parties such as bankers and lawyers, which expands the data exposure surface beyond the local planning task. In a skill handling business, governance, and capitalization information, such handoff instructions can lead an agent to prepare or disclose sensitive material without an explicit consent or review step.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
80% confidence
Finding

The entire skill reference document is written only in Chinese and does not indicate that users may choose another language or locale. Under the stated policy, forcing a specific language without user opt-in can be a natural-language policy violation unless the locale restriction is explicitly documented and justified.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
81% confidence
Finding

该参考文件全文仅以中文撰写,未说明这是面向特定中文用户群体的限定文档,也未提供其他语言或语言选择机制。根据规则,若技能内容强制单一语言而无用户选择或明确、合理的地域性说明,可构成自然语言层面的语言/locale 政策问题。

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
78% confidence
Finding

该参考文件全文以中文编写,且未见任何说明允许用户选择其他语言或地区化输出方式。按规则,若技能内容以自然语言强制单一语言而无用户选择或明确的地区合规理由,可构成自然语言政策问题。

Content

No source excerpt is available for this finding.