Back to skill

Security audit

Medical Device Reg Hub(全球医疗器械注册中枢)

Security checks for vulnerabilities and agentic risk

Overview

This documentation-only skill provides medical-device registration references, with minor configuration inconsistencies and outdated verification records.

Use this as a reference index for medical-device questions, with access limited to bundled documents. Verify current official requirements before compliance decisions. Regulatory accuracy, live link availability, and publisher assurances were not independently established; the bundled security records are incomplete and outdated.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (2)

The workflow requires reading local reference files despite declaring filesystem access disabled.

Medium
Category
Excessive Agency
Confidence
97% confidence
Finding

The manifest explicitly sets filesystem: false at L044, but L073 directs the agent to load a local references/ file; L125 repeats this as an explicit file-read instruction. Although reference lookup is appropriate for this knowledge hub, the prescribed workflow exceeds its declared permissions.

Content

No source excerpt is available for this finding.

Broad keyword triggers activate the skill without requiring medical-device regulatory context.

Medium
Category
Trigger Abuse
Confidence
96% confidence
Finding

L108 instructs the agent to load this skill whenever any listed term appears, including generic terms such as “技术文件” (technical documents), “标签” (labels), and “说明书” (instructions/manuals). These terms occur routinely outside medical-device registration, but the activation instructions do not require that domain context or provide exclusions.

Content

No source excerpt is available for this finding.