Back to skill

Security audit

全国招标采购信息总站-中国招标网

Security checks for vulnerabilities and agentic risk

Overview

This procurement-search skill needs review because its registration notice understates the privacy implications of recognizing your device.

Configure your own service API key to skip device collection, or request a clearer explanation of device-identifier retention and linkage before accepting automatic registration. Searches send your criteria to Zhiliaobiaoxun and may consume credits; package identity and notification-display status are also reported. Protect the saved key and auto-login links. All bundled files were reviewed, but the provider's actual server-side data practices are not established by these artifacts.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Rogue AgentSelf-Modification, Session Persistence
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (10)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The supplied code generates a device identifier rather than querying or analyzing procurement information. Local MAC-address access and device fingerprinting are not disclosed in the declared purpose, and the chunk provides no evidence that they are necessary supporting implementation details. No network transmission or exfiltration is shown.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The supplied code performs local device identification rather than procurement-information retrieval or analysis. Hardware-address collection and fingerprinting are not declared, and the snippet establishes no necessary connection to the described functionality. No transmission, storage, or registration is shown, so those behaviors are not inferred.

Content

No source excerpt is available for this finding.

The search workflow sends user-supplied company names and query criteria to a third-party API without a user-facing data-sharing warning.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

The workflow instructs the agent to POST a company name supplied for the user's query to the external Zhiliaobiaoxun API. Although the file documents API endpoints and requires consent for device collection during registration, it does not instruct the agent to disclose that normal searches transmit user-provided query information to this service.

Content

No source excerpt is available for this finding.

Automatic update checks and display acknowledgments transmit account-linked metadata without a user-facing privacy notice.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

The file instructs the agent to make an authenticated account request at the first use in each session and then submit an acknowledgment when an update notice is displayed. These requests transmit package identity, installation source, and notice-display status to the external service, but the skill provides no user-facing disclosure of this background reporting; stating that the requests are free does not explain the privacy implications.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
70% 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 · 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
60% 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 · 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

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

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**

Automatic registration collects and transmits a persistent hardware-derived identifier beyond the stated bidding-analysis purpose.

Medium
Category
Excessive Agency
Confidence
92% confidence
Finding

Although registration requires consent, the supplied privacy notice inaccurately characterizes a stable MAC-derived hash as having no identity implications and says no personal information is collected. A deterministic, unsalted MAC hash remains a persistent, linkable device identifier and is susceptible to candidate enumeration, so those assurances can undermine informed consent to hardware tracking.

Content

No source excerpt is available for this finding.

The skill mandates an unrelated package-update notification and remote acknowledgement workflow.

Low
Category
Excessive Agency
Confidence
90% confidence
Finding

The manifest describes bidding and procurement research, but these instructions require an account request at the first use in every session to check package updates, relay update messages, and submit a POST acknowledgement recording their display. This adds remote notification-state management unrelated to the stated research purpose, rather than merely authenticating or executing procurement queries.

Content

No source excerpt is available for this finding.

The skill mandates an unrelated package-update notification and display-tracking workflow.

Low
Category
Excessive Agency
Confidence
90% confidence
Finding

L351–L353 require an authenticated account request each session to check for package-update notices, relay the server-provided message, and submit a POST acknowledgment recording its display. This package-maintenance and notification-tracking capability is separate from the bidding searches, company analysis, and procurement intelligence described in the manifest; it is not merely authentication needed to perform those queries.

Content

No source excerpt is available for this finding.