Back to skill

Security audit

全网招中标数据平台-知了标讯

Security checks for vulnerabilities and agentic risk

Overview

The procurement features fit their purpose, but optional signup understates the privacy impact of its persistent device identifier.

Use a manually obtained API key to skip device fingerprinting. Before accepting automatic signup, clarify how the provider retains and handles its persistent device identifier. Searches send company names, keywords and filters to Zhiliaobiaoxun and may consume account credits. Protect the saved key and generated login links. All bundled files were inspected, but the provider's server-side privacy practices remain unverified.

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

声明的用途是招投标数据查询与分析,但此代码访问本机系统网络信息并生成设备标识,未体现与招投标查询有关的功能,也未说明设备识别是必要的支持步骤,因此构成未声明的能力及资源访问。该片段未显示向外发送数据或实际执行注册。

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The supplied code derives a device identifier rather than querying or analyzing procurement data. Local hardware identification is not disclosed in the description, and the chunk provides no evidence that it is necessary to support the declared functionality. The code does not itself transmit the identifier or perform registration.

Content

No source excerpt is available for this finding.

Automatic registration collects and transmits device fingerprints unrelated to bidding-data queries.

Medium
Category
Excessive Agency
Confidence
88% confidence
Finding

L054–L055 instruct the agent to collect platform, architecture, and a MAC-derived hash and transmit them to the registration endpoint. API authentication is expected for this skill, but hardware fingerprinting is not an obvious requirement of its stated procurement-data search and analysis purpose. The required consent mitigates covert collection but does not establish why device identification is necessary for that purpose.

Content

No source excerpt is available for this finding.

External transmission of user-supplied business queries lacks a privacy notice.

Medium
Category
Output Handling
Confidence
91% confidence
Finding

The query workflow sends the user's target company name, and analogous workflows send search keywords and filters, to the external Zhiliaobiaoxun API. Although the file documents API endpoints and requires consent for device-feature collection during registration, it does not instruct the agent to disclose that users' business research inputs will also leave their environment.

Content

No source excerpt is available for this finding.

Automatic registration collects and transmits a persistent hardware fingerprint beyond the bidding-analysis purpose.

Medium
Category
Excessive Agency
Confidence
89% confidence
Finding

The flow transmits a deterministic, unsalted MAC-address hash while describing the collected features as non-identifying and claiming that no personal information is collected. Hashing does not make a stable hardware identifier anonymous: it remains linkable across registrations, and a known MAC can be checked against its hash, so the consent explanation understates the privacy consequences.

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**

The skill requires an authenticated package-update notification and acknowledgment workflow unrelated to bidding analysis.

Low
Category
Excessive Agency
Confidence
88% confidence
Finding

L393–L395 require a first-use account request to check upgrade notices, forwarding the notice to the user, and an authenticated POST acknowledging its display. Package-update notification handling and modification of server-side notification state are not direct requirements of the manifest's procurement search and analysis functions, even though the instructions explicitly exclude purchases and charges.

Content

No source excerpt is available for this finding.

The skill sends an authenticated update-display receipt without notifying the user about this reporting.

Low
Category
Output Handling
Confidence
87% confidence
Finding

The workflow instructs the agent to POST a notice identifier and displayed: true to the API using the configured credentials after showing an update notice. This reports an account-linked interaction to the service, but the file provides no user-facing privacy disclosure and explicitly says not to display the technical receipt.

Content

No source excerpt is available for this finding.