Back to skill

Security audit

公众号情报分析系统

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed WeChat public-account monitoring and reporting tool that uses a local workspace and the Mangyun API, with no artifact-backed evidence of hidden exfiltration or destructive behavior.

Before installing, confirm you are comfortable giving the skill a Mangyun API key via MANGYUN_API_KEY, creating a local workspace of monitored article data, and making paid API calls within the configured budgets. Review the account list and output path before scanning or exporting, and do not point baseUrl at an untrusted service.

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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (20)

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

声明描述的是一个面向公众号情报分析系统的业务能力集合,但这段代码并未实现这些核心功能。它没有抓取公众号文章、监控多公众号、分析文章内容、生成简报或导出公众号情报数据。相反,代码集中在平台基础设施层:生成身份头、拼接产品 openapi 路径、解析标准响应错误、读取产品 catalog、按参数估算价格,以及将结果保存为快照并比较差异。虽然其中有对微信文章短链(mp.weixin.qq.com / weixin.qq.com)的价格调整判断,说明某些产品参数可能与微信链接有关,但这只是计费规则的一部分,不能等同于真正的公众号情报分析能力。因此,这段代码与声明用途存在实质性不匹配。

Content

No source excerpt is available for this finding.

Vague Triggers

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

Implicit invocation is enabled without any scoped guardrails, allowing the platform to trigger this skill automatically from general user phrasing. In this context, the skill advertises monitoring multiple accounts, article collection, AI analysis, reporting, and customer data export, so unintended invocation could lead to over-collection, privacy-impacting workflows, or execution of sensitive business-intelligence tasks the user did not explicitly authorize.

Content

No source excerpt is available for this finding.

Env Variable Harvesting

High
Category
Data Exfiltration
Confidence
60% confidence
Finding

Code enumerates, copies, or searches environment variables for secrets. Bulk environment access can collect credentials unrelated to the skill's stated purpose.

Content

Scanner excerpt · scripts/test_intelligence.py (reported line 119)May include surrounding context.

python
GatewayHandler.history_calls = 0
        self.temp = tempfile.TemporaryDirectory()
        self.workspace = Path(self.temp.name) / "workspace"
        self.env = os.environ.copy()
        self.env["MANGYUN_API_KEY"] = "test-key-never-persisted"
        self.env["MANGYUN_INTEL_TEST_ALLOW_HTTP"] = "1"
        self.env["PYTHONUTF8"] = "1"

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
94% confidence
Finding

The skill describes operations that inherently require filesystem, environment, network, and shell access, but it does not declare any explicit tool scope or permission boundaries. This weakens least-privilege controls and makes it easier for a host agent to grant broader capabilities than necessary, increasing the blast radius if the skill is misused or prompt-injected through untrusted article content.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The auto-trigger conditions are broad enough to overlap with ordinary discussion of monitoring, analysis, dashboards, or exports, which can cause the skill to activate without clear user intent. In a skill with network, filesystem, and shell-capable workflow steps, accidental activation can lead to unnecessary external requests, local data creation, or cost-incurring actions.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

L052 写明“使用简洁中文输出”,这是自然语言层面的固定语言要求,但文档中未见用户可选择其他语言或需中文的合规性说明。按照语言/地区政策,这属于未获用户选择的语言限制。

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 109)May include surrounding context.

先用小样本验证;未经我同意不自动扩大采集范围、不创建定时任务。

text

文档入口:https://api.we-media.cn/client/ai-guide
API 市场与逐产品 AI 文档:https://api.we-media.cn/client/products
公开产品目录:GET https://api.we-media.cn/api/v1/public/products
配置 Key 后,预估使用同 Key 请求 GET https://api.we-media.cn/api/v1/products,考虑参数附加价与调用次数;不能把价格写死。

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 110)May include surrounding context.

先用小样本验证;未经我同意不自动扩大采集范围、不创建定时任务。

text

文档入口:https://api.we-media.cn/client/ai-guide
API 市场与逐产品 AI 文档:https://api.we-media.cn/client/products
公开产品目录:GET https://api.we-media.cn/api/v1/public/products
配置 Key 后,预估使用同 Key 请求 GET https://api.we-media.cn/api/v1/products,考虑参数附加价与调用次数;不能把价格写死。

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 111)May include surrounding context.

先用小样本验证;未经我同意不自动扩大采集范围、不创建定时任务。

text

文档入口:https://api.we-media.cn/client/ai-guide
API 市场与逐产品 AI 文档:https://api.we-media.cn/client/products
公开产品目录:GET https://api.we-media.cn/api/v1/public/products
配置 Key 后,预估使用同 Key 请求 GET https://api.we-media.cn/api/v1/products,考虑参数附加价与调用次数;不能把价格写死。

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 112)May include surrounding context.

先用小样本验证;未经我同意不自动扩大采集范围、不创建定时任务。

text

文档入口:https://api.we-media.cn/client/ai-guide
API 市场与逐产品 AI 文档:https://api.we-media.cn/client/products
公开产品目录:GET https://api.we-media.cn/api/v1/public/products
配置 Key 后,预估使用同 Key 请求 GET https://api.we-media.cn/api/v1/products,考虑参数附加价与调用次数;不能把价格写死。

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The default prompt is a broad natural-language invocation that can activate the skill for a wide range of loosely related requests about public-account monitoring and analysis. Because the skill concerns large-scale content collection, analysis, dashboards, and data export, vague triggering increases the chance of unintended activation and data-processing actions outside the user's precise intent.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The schema and field content examples require Chinese-language summaries and text fields such as summary, logic, and angle, but the document does not provide user opt-in or an explicit justification for enforcing this locale. This is a natural-language policy concern because it constrains language behavior across the skill's outputs.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This Python file contains extensive hard-coded Chinese user-facing messages and CLI help text, indicating the skill operates in a fixed language/locale. The file does not offer any user opt-in, language selection, or documented locale justification, which matches the policy-violation criterion for forced language choice.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The code explicitly converts dates to Asia/Shanghai natural days and the generated dashboard HTML is fixed to zh-CN with Chinese UI text. This imposes a specific language and locale behavior without any visible opt-in or configuration path in the file.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The HTML document declares lang="zh-CN", and associated script formatting uses Chinese locale conventions and Asia/Shanghai timezone. This is a natural-language and locale policy issue because the skill forces a specific locale rather than allowing selection or clearly scoping the skill to that region.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The export_workbook function assembles detailed reporting data, including article content, account metadata, and API call records, then writes an XLSX file to the provided target path. There is no confirmation prompt, warning message, or user-visible disclosure in this file indicating that potentially sensitive operational data will be persisted to disk.

Content

No source excerpt is available for this finding.

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/test_intelligence.py (reported line 139)May include surrounding context.

python
self.temp.cleanup()

    def exec_cli(self, *args: str, expect: int = 0):
        result = subprocess.run(
            [sys.executable, str(SCRIPT), "--workspace", str(self.workspace), *args],
            env=self.env, text=True, encoding="utf-8", capture_output=True,
        )

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
76% confidence
Finding

The display name, description, and default prompt are all presented in Chinese, which can impose a language expectation without stating that the skill is Chinese-only or offering a language choice. This may conflict with language/locale policy expectations when users have not opted into that locale.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
87% confidence
Finding

文件标题及全文指令均以中文固定表达,未说明这是仅面向特定中文团队或区域的受限技能文档,也未提供语言选择或用户选择退出的说明。按照自然语言策略要求,强制特定语言而无选择机制可能构成语言/locale 策略违规。

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
96% confidence
Finding

The raised ValueError message at L107 is written only in Chinese, which imposes a specific language in user-facing output. The file does not offer a language choice or document a justified locale restriction, so this is a natural-language policy violation under the language/locale rule.

Content

No source excerpt is available for this finding.