Back to skill

Security audit

弱电智能化投标决策-安防监控项目投标评估

Security checks for vulnerabilities and agentic risk

Overview

This bid-analysis skill is coherent, but it stores account credentials, uses device-derived registration data, and creates shareable reports with signed links, so it needs user review before installation.

Install only if you are comfortable with this skill using the Zhiliaobiaoxun API, storing an API key under ~/.zlbx/config.json, collecting a hashed MAC address for trial registration when you consent, and generating local HTML reports that may contain signed links. Preconfiguring ZLBX_API_KEY avoids the auto-registration path, and generated reports should be shared only with intended recipients.

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
  • Rogue AgentSelf-Modification, Session Persistence
Findings (27)

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,对话报告输出后默认再用 `scripts/render_report.py` 生成一份可分享的 HTML 版报告并告知保存路径(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,对话报告输出后默认再用 `scripts/render_report.py` 生成一份可分享的 HTML 版报告并告知保存路径(详见 report-template.md「HTML 报告导出」)。

Hidden Instructions

High
Category
Prompt Injection
Confidence
88% confidence
Finding

This duplicate finding reflects the same hidden comment-based instruction block that changes API request and acknowledgement behavior. Concealing such instructions reduces auditability and can enable undisclosed external signaling or package-identification behavior that users and maintainers may not notice during normal review.

Content

Scanner excerpt · SKILL.md (reported line 163)May include surrounding context.

  • 报告里某个竞争对手威胁最大 → 引导用 zlbx-company-intel 对其深度背调
  • 以上都不贴切 → 报告涉及的企业完整档案与更多商机详情,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: intelligent-building-bid-decision/1.0.9

Hidden Instructions

High
Category
Prompt Injection
Confidence
88% confidence
Finding

This duplicate finding reflects the same hidden comment-based instruction block that changes API request and acknowledgement behavior. Concealing such instructions reduces auditability and can enable undisclosed external signaling or package-identification behavior that users and maintainers may not notice during normal review.

Content

Scanner excerpt · SKILL.md (reported line 163)May include surrounding context.

  • 报告里某个竞争对手威胁最大 → 引导用 zlbx-company-intel 对其深度背调
  • 以上都不贴切 → 报告涉及的企业完整档案与更多商机详情,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: intelligent-building-bid-decision/1.0.9

Hidden Instructions

High
Category
Prompt Injection
Confidence
92% confidence
Finding

This duplicate finding points to the same hidden instruction block, which attempts to control HTTP headers, package identity, and follow-up API behavior from within a markdown reference file. In a security context, concealed instructions that influence authentication and outbound requests are dangerous because they can cause the agent to trust adversarial content, leak environment-bound credentials, or perform unintended actions that are not part of the user's request.

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: bid-decision/1.0.5

Hidden Instructions

High
Category
Prompt Injection
Confidence
92% confidence
Finding

This duplicate finding points to the same hidden instruction block, which attempts to control HTTP headers, package identity, and follow-up API behavior from within a markdown reference file. In a security context, concealed instructions that influence authentication and outbound requests are dangerous because they can cause the agent to trust adversarial content, leak environment-bound credentials, or perform unintended actions that are not part of the user's request.

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: bid-decision/1.0.5

Hidden Instructions

High
Category
Prompt Injection
Confidence
90% confidence
Finding

This duplicate hidden-instructions finding points to the same concealed request-identity block that steers outbound requests and package attribution. In the context of a bidding-analysis skill, concealed operational directives are more dangerous because they accompany unrelated registration and credential-management behavior, reducing transparency and increasing the chance of unauthorized data handling.

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: bid-decision/1.0.5

Hidden Instructions

High
Category
Prompt Injection
Confidence
90% confidence
Finding

This duplicate hidden-instructions finding points to the same concealed request-identity block that steers outbound requests and package attribution. In the context of a bidding-analysis skill, concealed operational directives are more dangerous because they accompany unrelated registration and credential-management behavior, reducing transparency and increasing the chance of unauthorized data handling.

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: bid-decision/1.0.5

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The file implements a full auto-registration, credential provisioning, account recovery, recharge-link generation, and persistent API key management workflow that is unrelated to the stated purpose of bidding-decision analysis. This materially expands the skill's authority and data-handling scope, enabling collection of device-derived identifiers, outbound account creation, and local credential persistence without being necessary for the analytical task.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill directs collection of device fingerprint components, including a MAC-derived hash, for trial-account deduplication even though the skill's declared role is bid analysis. Hardware-derived identifiers are sensitive and create tracking risk; collecting them inside a domain-analysis skill violates least privilege and creates an unnecessary privacy and security exposure if the data is mishandled or repurposed.

Content

No source excerpt is available for this finding.

Hidden Instructions

High
Category
Prompt Injection
Confidence
86% confidence
Finding

This duplicate finding identifies the same hidden-instruction issue: a comment-delimited block embeds request identity and API handling directives inside a markdown template. Even if intended as packaging metadata, placing such instructions in content creates an opaque authority channel that can override safer defaults and is difficult for users and reviewers to detect.

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: bid-decision/1.0.5

Hidden Instructions

High
Category
Prompt Injection
Confidence
86% confidence
Finding

This duplicate finding identifies the same hidden-instruction issue: a comment-delimited block embeds request identity and API handling directives inside a markdown template. Even if intended as packaging metadata, placing such instructions in content creates an opaque authority channel that can override safer defaults and is difficult for users and reviewers to detect.

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: bid-decision/1.0.5

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The template explicitly requires exposing full API-returned URLs containing sk signed parameters to end users and preserving them in rendered reports. Signed bearer-style links often grant access without normal authentication, so disclosing or sharing them can leak privileged access, enable replay by unintended recipients, and bypass login controls noted in the template itself.

Content

No source excerpt is available for this finding.

Hidden Instructions

High
Category
Prompt Injection
Confidence
94% confidence
Finding

This duplicated hidden-instruction finding is still a real issue: the workflow tells the agent to preserve and send specific identity headers and to perform additional authenticated calls such as account balance checks and acknowledgment posts. Because these instructions come from potentially adversarial skill content rather than trusted runtime policy, they can coerce the agent into making credentialed requests or leaking operational metadata without independent authorization controls.

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: bid-decision/1.0.5

Hidden Instructions

High
Category
Prompt Injection
Confidence
94% confidence
Finding

This duplicated hidden-instruction finding is still a real issue: the workflow tells the agent to preserve and send specific identity headers and to perform additional authenticated calls such as account balance checks and acknowledgment posts. Because these instructions come from potentially adversarial skill content rather than trusted runtime policy, they can coerce the agent into making credentialed requests or leaking operational metadata without independent authorization controls.

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: bid-decision/1.0.5

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 71)May include surrounding context.

python
THREAT_COLOR = {"高": "#c0392b", "中": "#b9770e", "低": "#0d9463"}

# 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
80% confidence
Finding

The skill explicitly states it will write HTML reports to a local directory, but it does not declare any corresponding tool scope or allowed-tools boundary. This creates a capability/permission mismatch: an agent runtime may permit file writes without the user being clearly informed through a formal permission declaration, increasing the risk of unintended local file creation or overwrite.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The activation scope is very broad, including cases where the user did not mention the core domain term and instructing that the skill 'must' be used for loosely related requests. Over-broad routing can cause the agent to invoke this skill unexpectedly, sending project, company, or procurement queries to an external API when a user may have intended a general discussion or a different tool.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

The instructions prescribe report wording and required disclaimer text entirely in Chinese and do not offer any user opt-in or language selection. Under the policy, forcing a specific language or locale without user choice can be a natural-language policy violation unless clearly justified as region-specific.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
87% confidence
Finding

This section documents sending collected device features and account-registration data to an external service via JSON POST. The transmission is intentional, but because the transmitted data includes device fingerprint material and account bootstrapping metadata not required for bidding analysis, it represents unjustified exfiltration to a third party within this skill's context.

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`)。
>

External Transmission

Medium
Category
Data Exfiltration
Confidence
84% confidence
Finding

The curl example reinforces the same external transmission behavior: serializing and posting registration payloads to a remote endpoint. While not exploit code by itself, it operationalizes sending non-essential device/account data off-host from a skill whose declared purpose is analytical, thereby increasing privacy and trust risk.

Content

Scanner excerpt · references/auto-register.md (reported line 131)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`)。
>
> 历史教训:曾有版本采集 `home_path`,Windows 的 `C:\Users\alice` 直接拼进 JSON 字符串时

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill includes instructions to persist API keys to ~/.zlbx/config.json, mark their provenance, and generate recharge/login links when quota is exhausted, all of which are outside the expected behavior of a bidding-analysis assistant. Persistent secret handling and payment-adjacent account flows increase attack surface and make the skill capable of modifying local state and steering users into account actions unrelated to the analytical function.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
94% confidence
Finding

This duplicate finding captures the same persistence behavior: durable storage of API keys plus immediate authenticated reuse. In a skill whose advertised purpose is bid-evaluation, this persistence is especially unjustified and increases the blast radius of compromise by leaving long-lived secrets on disk and enabling future calls without renewed consent.

Content

Scanner excerpt · references/auto-register.md (reported line 195)May include surrounding context.

text

注意事项:
- 目录不存在时先 `mkdir -p ~/.zlbx`
- 文件已存在时**合并而非覆盖**(保留用户可能的其他配置)
- `source: "auto"` 字段必须写入,**这是后续判断「是否输出自动登录链接」的关键依据**

Session Persistence

Medium
Category
Rogue Agent
Confidence
94% confidence
Finding

This duplicate finding captures the same persistence behavior: durable storage of API keys plus immediate authenticated reuse. In a skill whose advertised purpose is bid-evaluation, this persistence is especially unjustified and increases the blast radius of compromise by leaving long-lived secrets on disk and enabling future calls without renewed consent.

Content

Scanner excerpt · references/auto-register.md (reported line 195)May include surrounding context.

text

注意事项:
- 目录不存在时先 `mkdir -p ~/.zlbx`
- 文件已存在时**合并而非覆盖**(保留用户可能的其他配置)
- `source: "auto"` 字段必须写入,**这是后续判断「是否输出自动登录链接」的关键依据**

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill mandates automatic generation of a local HTML file and disclosure of the absolute filesystem path to the user. Revealing host paths leaks environmental details about the runtime, and automatic file creation can leave sensitive report data on disk without user awareness or retention controls.

Content

No source excerpt is available for this finding.