Back to skill

Security audit

标讯 · 广东招标中标与招投标数据查询

Security checks for vulnerabilities and agentic risk

Overview

The skill’s procurement queries and optional account setup are purpose-aligned, with consent required before collecting device details.

Use an existing API key or manual signup if you prefer to avoid device fingerprinting. Automatic signup requires consent and sends a stable MAC hash that can link registrations to your device. Queries consume service credits, and supplemental web searches share research terms externally. Keep confidential terms out of searches and protect the saved API key and temporary login links. The artifacts do not establish how the provider retains or reuses submitted data.

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 (7)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The supplied code performs local device identification rather than procurement data retrieval or analysis. Hardware-derived fingerprinting is not disclosed in the declared purpose, and the chunk shows no connection that would establish it as a necessary supporting implementation detail. The code outputs the hash; it does not show transmission, storage, or registration.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The supplied command performs local device fingerprinting rather than procurement-related querying or analysis. The declared purpose does not disclose device identification or explain why access to a host MAC address is necessary. Although the file name suggests registration, the code itself establishes no connection to a declared feature. No transmission, persistence, or account registration is shown in this chunk.

Content

No source excerpt is available for this finding.

Supplemental WebSearch is recommended without warning that user-derived research terms will be sent to an external search service.

Medium
Category
Output Handling
Confidence
87% confidence
Finding

The skill recommends using WebSearch for company, competitive, and market analysis as part of its own workflow, rather than merely listing an available tool. It does not instruct the agent to disclose that company names or research keywords derived from the user's request will be transmitted to an additional service, potentially revealing confidential business interests.

Content

No source excerpt is available for this finding.

Automatic registration collects and transmits a persistent hardware fingerprint not justified by the manifest’s tender-query and analysis purpose.

Medium
Category
Excessive Agency
Confidence
92% confidence
Finding

The workflow reads a network adapter’s MAC address, computes an unsalted SHA-256 hash, and sends it with operating-system and architecture information to an account-registration endpoint. Hardware-based trial-account deduplication is an additional device-identification capability, not an obvious requirement of tender searches, business analysis, or balance queries declared in the manifest. The explicit consent gate mitigates undisclosed collection but does not make this capability part of the stated scope.

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**