Back to skill

Security audit

公允价值·逐条挂源

Security checks for vulnerabilities and agentic risk

Overview

The scanners report a suspicious skill because its financial disclosure purpose does not clearly explain its online MCP discovery and credential workflow, alongside concerns about external document parsing and checker coverage; the scan was partial.

Consider avoiding installation until the online MCP workflow, credential destination, and external document parsing are clearly explained and limited to destinations you trust. The scan is partial: 3 of 4 components were scanned, and static analysis of a shell script was degraded, so the evidence is incomplete.

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 (8)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

该代码没有读取财报或公告、提取公允价值披露或建立原句引用登记表,而是获取并打印外部 MCP 目录。所访问资源及实际执行任务与声明用途没有可见关联,因此构成实质性不匹配。代码未显示上传本地数据或执行下载内容。

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

所示代码的实际功能是发现 MCP 服务及列举远程工具,未体现财报读取、公允价值披露提取、原句引用或锚点登记。代码也未说明该远程服务与声明任务有何必要关联,因此不能将这些网络访问及认证操作视为已明确对应声明用途的辅助实现。另需注意,代码只打印目录中的 routing,并未给 ROUTING_URL 赋值;后续认证请求的实际目标取决于已有环境配置。

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
90% confidence
Finding

The anonymous catalog GET is not itself evidence of sensitive-data exfiltration, but the surrounding workflow directs the user to send CUE_API_KEY to a routing URL obtained from that remote catalog. Without an independent destination restriction, a compromised or manipulated routing response could cause the bearer credential to be sent to an attacker-controlled endpoint.

Content

Scanner excerpt · README.md (reported line 86)May include surrounding context.

发现先行,匿名取 live 目录,拿到 routing 后再连接发 tools/list,以现场返回为准,不硬编码工具名与工具数:

bash
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.get('routing') or d)"
# 拿到 routing 后,带 token 发 tools/list:
curl -sS -X POST "$ROUTING_URL" -H "Authorization: Bearer $CUE_API_KEY" \

The documented data-channel commands send HTTP requests despite the explicit statement that the skill makes no requests.

Medium
Category
Excessive Agency
Confidence
88% confidence
Finding

The README instructs users to make external requests, including an authenticated request, but later states that the skill makes no requests. Although these are documented commands rather than automatically executed code, the contradictory assurance can undermine informed authorization for network access and credential use.

Content

No source excerpt is available for this finding.

The document-parsing workflow lacks a warning that attachments may be transmitted to an external service.

Medium
Category
Output Handling
Confidence
94% confidence
Finding

The workflow directs long attachments and scans to an external parsing service without explaining the document-disclosure implications or requiring consent. Public annual reports reduce the typical risk, but user-supplied files and annotations may contain confidential information that should not leave the user's environment.

Content

No source excerpt is available for this finding.

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 workflow sends attachments to an external parsing service without a privacy warning.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

The skill directs external parsing when evidence resides in attachments, but it does not require checking document sensitivity or obtaining permission for that disclosure. Financial reports are often public, yet the accepted user-supplied screenshots and files can include nonpublic information or annotations.

Content

No source excerpt is available for this finding.

The documented whole-document red-line check excludes the gap section and section headings.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The checker advertises whole-document screening for valuation opinions and ratings, but check_line skips the entire gap section and _all_lines omits section headings. A crafted draft can therefore place prohibited assertions in those locations and still receive a successful result, undermining the integrity of downstream reliance on that result.

Content

No source excerpt is available for this finding.