Back to skill

Security audit

工商数据查询 · 企业基本信息、股东、分支、投资与变更

Security checks for vulnerabilities and agentic risk

Overview

This company-information skill discloses its paid queries, requires consent for device-based registration, and shows no material security concern.

Expect provider charges for successful queries and protect the saved API key and signed login links. Automatic registration asks permission to send a persistent device hash; configuring an existing key skips that collection. All bundled files and scanner coverage exceptions were reviewed, but this review does not verify the provider's backend or data-retention claims.

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
  • Rogue AgentSelf-Modification, Session Persistence
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (6)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
85% confidence
Finding

The declared enterprise-query functionality does not disclose collection or hashing of local hardware identifiers. Although the file location suggests a connection to automatic registration, the supplied code only generates a device fingerprint and does not establish that this access is necessary for the declared registration flow. No transmission, registration, or consent handling is shown.

Content

No source excerpt is available for this finding.

The workflow adds paid bidding intelligence and contact lookups beyond the manifest's declared company-registry capabilities.

Medium
Category
Excessive Agency
Confidence
92% confidence
Finding

The manifest at L003 describes ten registry-related capabilities and directs broader company intelligence to another skill. However, L110–126 instruct this skill to call bidding-profile, project-contact, bid-history, and customer/supplier tools, including additional 1- and 5-point charges. These are substantive additional data services rather than implementation details of registry queries; requiring a user request limits unexpected execution but does not resolve the manifest mismatch.

Content

No source excerpt is available for this finding.

The skill authorizes paid bidding and contact queries beyond its declared company-registry scope.

Medium
Category
Excessive Agency
Confidence
93% confidence
Finding

L130–L140 authorize supplementary calls for bidding profiles, procurement contacts, award records, and customer/supplier relationships. These are additional paid capabilities beyond the manifest's ten registry-related functions and its routing of broader company intelligence and project decisions to other skills; making them optional does not resolve the scope mismatch.

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**