Back to skill

Security audit

保险销售合规自查顾问

Security checks for vulnerabilities and agentic risk

Overview

This local compliance checklist has proportionate access, but a confirmed calculation error can understate its risk ratings.

Use this as a checklist with qualified human review. Have the scoring error corrected before relying on its ratings, verify applicable rules independently, and save completed CSV records elsewhere before exporting again. Customer records need not be supplied to the script.

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
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • 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
Findings (2)

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: file_write.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

The double-recording control adjustment reduces every risk group, contradicting the comment that it reduces group A risk.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

Selecting --douluo yes multiplies every group's exposure by 0.78, although the stated control concerns double-recording and group A already receives a separate reduction. This understates unrelated risks, such as premium handling and customer-data protection, and can lower the overall risk classification without evidence of improved controls in those areas. In a compliance self-check tool, that creates misleading decision support despite the human-review disclaimer.

Content

No source excerpt is available for this finding.