Back to skill

Security audit

医疗设备商机雷达-医院采购早期发现

Security checks for vulnerabilities and agentic risk

Overview

The skill mostly matches its procurement-opportunity purpose, but it also performs sensitive onboarding, device fingerprinting, local credential persistence, and bearer-style link sharing that deserve manual review.

Install only if you are comfortable with this third-party service receiving your search terms and, if you use auto-registration, device-derived identifiers. Prefer supplying your own ZLBX_API_KEY through a trusted secret mechanism, avoid sharing generated reports or chat logs that contain sk links, and review ~/.zlbx/config.json if you later want to remove stored credentials.

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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Findings (35)

Vague Triggers

High
Category
Not specified by scanner
Confidence
96% confidence
Finding

The trigger condition is overly broad and instructs mandatory use of this skill even when the user did not explicitly mention medical context, as long as the request vaguely relates to hospital procurement or sales leads. This can cause inappropriate routing of user queries to an external service, leading to unnecessary disclosure of user intent, over-collection, and actions the user did not clearly request.

Content

No source excerpt is available for this finding.

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 81)May include surrounding context.

md
完整模式下,对话清单输出后默认再用 `scripts/render_report.py` 生成一份可分享的 HTML 版商机清单并告知保存路径(详见 report-template.md「HTML 报告导出」)。

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 85)May include surrounding context.

md
完整模式下,对话清单输出后默认再用 `scripts/render_report.py` 生成一份可分享的 HTML 版商机清单并告知保存路径(详见 report-template.md「HTML 报告导出」)。

Hidden Instructions

High
Category
Prompt Injection
Confidence
95% confidence
Finding

This duplicate finding identifies the same hidden-instructions block, which contains non-obvious directives for request headers and post-display acknowledgement behavior. Because the hidden block governs outbound request metadata and follow-up network calls, it creates transparency and auditability risks beyond normal documentation comments.

Content

Scanner excerpt · SKILL.md (reported line 171)May include surrounding context.

  • 用户想持续跟踪 → 介绍订阅模式(见上方「订阅模式」)
  • 以上都不贴切 → 询问是否要对某条高价值商机深挖,或引导访问知了商机大师 https://agent.zhiliaobiaoxun.com

安装包身份与升级检查:所有发往知了标讯 API 的请求(包括账户查询)必须额外携带以下两个请求头:

text
X-Skill-Package: medical-equipment-opportunity-radar/1.0.8

Hidden Instructions

High
Category
Prompt Injection
Confidence
95% confidence
Finding

This duplicate finding identifies the same hidden-instructions block, which contains non-obvious directives for request headers and post-display acknowledgement behavior. Because the hidden block governs outbound request metadata and follow-up network calls, it creates transparency and auditability risks beyond normal documentation comments.

Content

Scanner excerpt · SKILL.md (reported line 171)May include surrounding context.

  • 用户想持续跟踪 → 介绍订阅模式(见上方「订阅模式」)
  • 以上都不贴切 → 询问是否要对某条高价值商机深挖,或引导访问知了商机大师 https://agent.zhiliaobiaoxun.com

安装包身份与升级检查:所有发往知了标讯 API 的请求(包括账户查询)必须额外携带以下两个请求头:

text
X-Skill-Package: medical-equipment-opportunity-radar/1.0.8

Hidden Instructions

High
Category
Prompt Injection
Confidence
96% confidence
Finding

This duplicated hidden-instruction finding is substantively valid for the same reason: the skill uses concealed prompt content to control authentication-adjacent request headers and provenance. In an adversarial-review model, concealed behavioral directives are a red flag because they bypass normal user visibility and can alter how external services trust the agent.

Content

Scanner excerpt · references/auto-register.md (reported line 1)May include surrounding context.

请求身份:使用已配置的 API Key,并保留本包的 X-Client,携带下列完整身份头;不得沿用其它包的版本或来源。

text
X-Client: opportunity-radar/1.0.3

Hidden Instructions

High
Category
Prompt Injection
Confidence
96% confidence
Finding

This duplicated hidden-instruction finding is substantively valid for the same reason: the skill uses concealed prompt content to control authentication-adjacent request headers and provenance. In an adversarial-review model, concealed behavioral directives are a red flag because they bypass normal user visibility and can alter how external services trust the agent.

Content

Scanner excerpt · references/auto-register.md (reported line 1)May include surrounding context.

请求身份:使用已配置的 API Key,并保留本包的 X-Client,携带下列完整身份头;不得沿用其它包的版本或来源。

text
X-Client: opportunity-radar/1.0.3

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The file defines a full account lifecycle unrelated to the advertised medical-opportunity discovery function: automatic registration, API-key issuance, local credential persistence, and recharge/login flows. This expands the skill's authority far beyond its stated purpose and creates an unnecessary trust boundary where the skill can enroll users into a third-party service and manage credentials on their behalf.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The skill instructs collection of device fingerprinting data (platform, CPU architecture, and a MAC-derived hash) and transmission to an external service for account creation. Even with hashing, this is persistent cross-session device identification unrelated to the user-facing business purpose, increasing privacy risk and enabling tracking or duplicate-account controls outside user expectation.

Content

No source excerpt is available for this finding.

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · references/api-quick.md (reported line 1)May include surrounding context.

请求身份:使用已配置的 API Key,并保留本包的 X-Client,携带下列完整身份头;不得沿用其它包的版本或来源。

text
X-Client: opportunity-radar/1.0.3

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · references/report-template.md (reported line 1)May include surrounding context.

请求身份:使用已配置的 API Key,并保留本包的 X-Client,携带下列完整身份头;不得沿用其它包的版本或来源。

text
X-Client: opportunity-radar/1.0.3

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · references/api-quick.md (reported line 1)May include surrounding context.

请求身份:使用已配置的 API Key,并保留本包的 X-Client,携带下列完整身份头;不得沿用其它包的版本或来源。

text
X-Client: opportunity-radar/1.0.3

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · references/report-template.md (reported line 1)May include surrounding context.

请求身份:使用已配置的 API Key,并保留本包的 X-Client,携带下列完整身份头;不得沿用其它包的版本或来源。

text
X-Client: opportunity-radar/1.0.3

Ssd 3

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The template explicitly requires redisclosing full API-returned URLs containing an sk login-bypass parameter and forbids removing or modifying that token. Sharing bearer-style authenticated links in chat or exported reports can expose access tokens to unintended recipients, enable unauthorized access, and propagate credentialed URLs into logs, transcripts, or shared files.

Content

No source excerpt is available for this finding.

Hidden Instructions

High
Category
Prompt Injection
Confidence
94% confidence
Finding

This duplicated hidden-instructions finding identifies the same underlying issue: concealed package-authentication guidance embedded in the skill content. Even if intended for normal package routing, placing these directives in hidden markup makes them adversarial from the agent's perspective and can lead to silent authenticated requests or policy conflicts.

Content

Scanner excerpt · references/workflow.md (reported line 1)May include surrounding context.

请求身份:使用已配置的 API Key,并保留本包的 X-Client,携带下列完整身份头;不得沿用其它包的版本或来源。

text
X-Client: opportunity-radar/1.0.3

Hidden Instructions

High
Category
Prompt Injection
Confidence
94% confidence
Finding

This duplicated hidden-instructions finding identifies the same underlying issue: concealed package-authentication guidance embedded in the skill content. Even if intended for normal package routing, placing these directives in hidden markup makes them adversarial from the agent's perspective and can lead to silent authenticated requests or policy conflicts.

Content

Scanner excerpt · references/workflow.md (reported line 1)May include surrounding context.

请求身份:使用已配置的 API Key,并保留本包的 X-Client,携带下列完整身份头;不得沿用其它包的版本或来源。

text
X-Client: opportunity-radar/1.0.3

Obfuscated Code

High
Category
Supply Chain
Confidence
50% confidence
Finding

Code contains obfuscation (base64, hex encoding with execution). This is often used to hide malicious functionality.

Content

Scanner excerpt · scripts/render_report.py (reported line 64)May include surrounding context.

python
ROUTE_COLOR = {"拟建": "#7c5cbf", "意向": "#0b7a6a", "临期": "#b9770e"}

# 知了标讯白色 logo(299x96 PNG base64 内嵌,保证报告离线/打印/转发时不裂图)
_LOGO_B64 = "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
...[truncated 27 chars]

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
91% confidence
Finding

The skill explicitly writes HTML reports to a local directory (~/zlbx-opportunity-radar-files/) but does not declare any tool scope or permission boundary for file write access. Undeclared write capability weakens user consent and platform enforcement, increasing the risk of unexpected local file creation or persistence beyond what a user understood from installation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

L003 使用“必须使用此SKILL”并整体以中文规定交互方式,但未说明可根据用户语言偏好切换,也未提供语言/locale 选择。按规则,强制单一语言而无用户 opt-in 属于自然语言层面的语言政策问题。

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The skill includes an automatic account-registration flow and collects device-derived identifiers (platform, arch, mac_hash) when no API key is present, which exceeds the core function of procurement opportunity discovery. Even with stated consent language, this expands data collection and external side effects, creating privacy and trust risks if the user expects a search/reporting skill rather than account provisioning and fingerprint-based deduplication.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The documentation explicitly states that returned URLs include an sk parameter that enables direct access without login and instructs operators to output them verbatim. If those links are exposed to unintended recipients, logged in chat history, analytics, or shared externally, they can function as bearer-style access tokens and grant unauthorized access to sensitive procurement records. In a skill focused on hospital and government procurement intelligence, this is especially risky because the data may be commercially sensitive and may include downstream contact or contract context.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The file notes that expiring-project results contain contact fields but provides no concrete privacy or data-minimization warning beyond a vague instruction to be restrained. This can lead an agent to surface personal contact information from hospital or public-sector procurement records to users who may not need it, increasing privacy, compliance, and misuse risk. Because the skill is designed to generate sales opportunities, the context raises the chance that contact details would be used for unsolicited outreach.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The document instructs reading API keys from environment/config and writing new credentials into ~/.zlbx/config.json, establishing local credential persistence unrelated to the core medical-opportunity use case. This broadens the skill from data retrieval into local secret management, which increases the chance of insecure storage, unintended overwrites, and credential misuse.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

This markdown file contains mandated user-visible text in Chinese for consent and operational messaging, but does not indicate that the user may choose another language. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
95% confidence
Finding

This section explicitly instructs sending collected device features in a JSON POST request to an external domain. Because the transmitted fields include persistent device-identifying material and are used to create or recover service accounts, the outbound transfer is security-relevant and more dangerous in this skill context than ordinary API use.

Content

Scanner excerpt · references/auto-register.md (reported line 130)May include surrounding context.

md
> ### ⚠️ 请求体必须用 JSON 序列化函数生成,不要手拼字符串
>
> 用 `json.dumps(payload)` / `requests.post(url, json=payload)` / `JSON.stringify(payload)`,
> 或 `curl -d @file`;**不要用字符串拼接,也不要用 Python 的 `str(dict)`**
> (后者产出单引号,服务端会报 `Expecting property name enclosed in double quotes`)。
>