Back to skill

Security audit

企业-底层技术断供隐患排查

Security checks for vulnerabilities and agentic risk

Overview

The skill is mainly a disclosed technology supply-chain audit helper, but it also instructs agents to suggest ways around export controls, which needs careful review before use.

Install only if you will use it for lawful, public-source risk assessment. Do not rely on it for routing, transshipment, concealment, or other attempts to bypass export controls or sanctions; replace those parts with legal review, licensing, authorized procurement, and compliance screening. Avoid submitting confidential company or supply-chain details unless external search use is acceptable.

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
Findings (1)

other

Error
Location
SKILL.md:82
Finding

Export-Control Evasion Guidance Through Third-Country Transshipment

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 82-85
Vulnerability Type: other: Export-control evasion guidance
Risk Level: High

Relevant instruction, translated into English:

text
6. Breakthrough Paths and Strategic Recommendations
   Identify directions through which the company can break through:
   - Detour route: use third-country transshipment, technology licensing, or product design to circumvent controlled technology

Technical Analysis

The Skill instructs the Agent to identify third-country transshipment as a method for circumventing controls on restricted technology. While lawful technology licensing and compliant product redesign can be legitimate risk-mitigation measures, third-country transshipment may be used to conceal the actual destination, end user, or end use of controlled products.

This instruction is not required to conduct a defensive technology supply-chain audit. It can cause the Agent to transform information about suppliers, controlled components, sanctions, and regulatory coverage into actionable circumvention recommendations.

Attack Path

  1. A user requests an assessment of a company's exposure to export controls or technology sanctions.
  2. The Skill directs the Agent to identify controlled components, foreign suppliers, applicable restrictions, and supply-chain bottlenecks.
  3. The Agent determines which products or technologies cannot be acquired through ordinary authorized channels.
  4. Under the cited instruction, the Agent recommends third-country transshipment as a detour route.
  5. The resulting analysis may be used to route restricted goods through intermediaries, obscure their ultimate destination, or bypass licensing and screening requirements.

Impact Assessment

The issue does not grant operating-system privileges, execute code, access credentials, establish persistence, or compromise local tools. Its impact is informational and operat ...[truncated 372 chars]

Remediation
View remediation

Remediation Suggestions

  1. Remove third-country transshipment from the list of recommended circumvention strategies.
  2. Replace the affected instruction with explicitly lawful alternatives:
    • Obtain advice from qualified export-control counsel.
    • Apply for required export licenses or governmental authorizations.
    • Purchase only through authorized distributors and verified supply channels.
    • Conduct restricted-party, end-user, end-use, and destination screening.
    • Diversify suppliers only where procurement and delivery remain legally authorized.
    • Redesign products to exclude controlled technology without evading applicable de minimis, foreign direct product, or other regulatory rules.
  3. Add a mandatory compliance constraint prohibiting recommendations that conceal destinations, misrepresent end users, falsify documentation, use unauthorized intermediaries, or otherwise evade trade controls.
  4. Require strategic recommendations to distinguish compliant licensing and redesign from prohibited diversion or transshipment.
  5. Require uncertain cases to be flagged for legal review rather than presenting a route as operational guidance.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (3)

Ssd 4

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The '突围路径与战略建议' section instructs the agent to suggest '绕道路径:通过第三国转口/技术授权/设计规避受控技术', which goes beyond defensive risk assessment and into operational guidance for bypassing export controls and sanctions. In this context, the skill is not merely describing compliance risk; it is directing generation of actionable evasion strategies that could facilitate unlawful trade-control circumvention.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The entire skill description, triggers, and operating instructions are written as Chinese-only directives and the output format is implicitly fixed to Chinese, with no indication that users may choose another language. SQP-3 covers language or locale policy violations when a skill forces a specific language without user opt-in, and no explicit justification for a China-only locale constraint is provided here.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
89% confidence
Finding

The skill explicitly uses live web search and analyzes sensitive target-company supply-chain and sanctions exposure, but it does not warn users that company identifiers and investigative prompts may be sent to external tools. This creates a transparency and privacy-risk issue, especially when users may input non-public or commercially sensitive targeting context.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.