Back to skill

Security audit

标讯 · IT信息化招标中标与招投标数据查询

Security checks for vulnerabilities and agentic risk

Overview

The procurement tools fit their purpose, but conflicting registration instructions could cause unnecessary device sharing or replacement of an existing account key.

Prefer configuring your existing API key manually. Authentication failures should lead to credential recovery, with your current key preserved unless you approve replacement. Optional trial registration sends operating-system type, CPU architecture, and a persistent MAC-derived hash to the provider; hashing does not make that identifier anonymous. Queries also share search terms with the provider and may consume credits. All packaged files and scanner gaps were reviewed; live service behavior and privacy claims 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 (8)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The declared purpose does not disclose device identification or local hardware-address collection. The supplied command performs device fingerprinting rather than procurement research or account-balance retrieval, and no connection establishing it as a necessary supporting implementation detail is shown. The chunk only prints the hash; it does not demonstrate transmission, persistence, or account registration.

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 search or analysis. Reading and hashing a network interface's MAC address accesses a resource unrelated to the stated functionality, and the description does not disclose device fingerprinting. Although the filename suggests automatic registration, the snippet itself only computes and outputs the hash; it does not demonstrate registration, transmission, or storage.

Content

No source excerpt is available for this finding.

Authentication-error handling directs re-registration even when credentials already exist, contradicting the missing-credentials-only registration boundary.

Medium
Category
Excessive Agency
Confidence
96% confidence
Finding

The error table directs re-registration for invalid, expired, or disabled credentials, including when a key already exists, contradicting the missing-credentials-only registration gate. An agent following that branch could initiate unnecessary device collection or create a replacement account and change account context, although the separate consent and recovery requirements reduce the risk.

Content

No source excerpt is available for this finding.

Automatic company lookup sends user-provided company names to a third-party API without a user-facing data-sharing notice.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

The workflow instructs the agent to call search_company with the company name from the user's request and automatically use the results in subsequent queries. Unlike a passive API catalog entry, this is an instruction to transmit user-provided data, but it does not warn the user that their query information will be sent to 知了标讯.

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 fingerprint beyond the tender-query skill's stated purpose.

Medium
Category
Excessive Agency
Confidence
89% confidence
Finding

Although the prose requires consent, the minimal pseudocode proceeds directly from missing credentials to device-feature collection and registration without a consent check. An agent implementing that example could transmit a persistent MAC-derived identifier without authorization; hashing makes it pseudonymous, not inherently anonymous as the privacy wording suggests.

Content

No source excerpt is available for this finding.