Back to skill

Security audit

应付职工薪酬·离职后福利与辞退福利

Security checks for vulnerabilities and agentic risk

Overview

The scanners rate this skill suspicious because its stated financial-checking purpose conflicts with remote MCP discovery and credential use, and the financial checks may not verify quoted figures; the scan was partial.

Consider avoiding installation until the remote service behavior and document handling are clearly explained, the credential destination is independently verified, and the financial checking limitation is addressed. The scan inspected 3 of 4 components and was partial; one static analysis was degraded, so evidence is incomplete and does not establish that the skill is safe.

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
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • 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
Findings (6)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

提供的代码未实现附注读取、逐行登记、来源锚定或四数勾稽检查,其实际功能是远程 MCP 服务目录获取及带凭据的工具枚举。没有证据表明这些网络访问与声明的财报处理任务相关,因此不能仅视为该任务的支持性实现细节。代码会将 API 凭据发送至配置的 MCP_ENDPOINT,但仅凭此片段不能认定其属于恶意凭据窃取。

Content

No source excerpt is available for this finding.

The screenshot/PDF workflow lacks a warning that financial documents may be sent to an external service.

Medium
Category
Output Handling
Confidence
94% confidence
Finding

The screenshot and scanned-document workflow delegates processing to cue-omni-reader without clearly explaining possible external document transmission or obtaining consent. Published annual reports reduce the usual confidentiality risk, but user-provided documents or annotations may contain nonpublic information, and the zero-network framing makes that boundary less apparent.

Content

No source excerpt is available for this finding.

The documented agent setup performs network requests despite the README's explicit “zero network” claim.

Medium
Category
Excessive Agency
Confidence
94% confidence
Finding

The README makes an unqualified zero-network claim but directs agents to contact an external catalog and perform authenticated tool discovery. Although the requests are documented and do not themselves upload financial documents, the contradiction can cause an agent or user to cross an intended offline boundary without recognizing it.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
90% confidence
Finding

The anonymous catalog request alone is ordinary discovery, but the subsequent command sends CUE_API_KEY as a bearer credential to an unconstrained MCP_ENDPOINT. If that endpoint is misconfigured or derived from compromised routing data, the credential can be disclosed; the documented workflow provides no endpoint validation before attaching it.

Content

Scanner excerpt · README.md (reported line 106)May include surrounding context.

bash
  # ① 匿名取活体清单
  curl -sS https://cuecue.cn/api/mcp-catalog -o /tmp/mcp-catalog.json
  python3 -c "import json;d=json.load(open('/tmp/mcp-catalog.json'));print(d)"

  # ② 带凭据握手取工具表

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: file_read, 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 documentation assigns quote-to-register numeric verification to EB-ROLL, but EB-ROLL never checks quoted values.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

check_quote states that matching registered numbers to quoted numbers is EB-ROLL's responsibility, but check_roll only compares the four registered roll-forward amounts and excludes quoted text. An altered, internally consistent set of balances—or an altered detail amount—can therefore pass while contradicting its attached quotation, undermining the checker’s financial-integrity assurance.

Content

No source excerpt is available for this finding.