Back to skill

Security audit

知了标讯官方招投标助手

Security checks for vulnerabilities and agentic risk

Overview

This procurement assistant needs review because conflicting login-error instructions could trigger unintended device registration and account changes.

Resolve the conflicting re-registration instructions before use: existing-key failures should lead to user-directed recovery. Providing your own API key normally skips device registration. Searches reach Zhiliao, and optional trial registration sends a stable device hash. Keep keys and login links private. This review covered every bundled file, but did not verify the hosted service's behavior or data retention.

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 concerns procurement-data queries and analysis, not host identification or network-interface inspection. The supplied code instead produces a hash derived from a local MAC address, with no demonstrated connection to the declared functions. No transmission, persistence, or actual registration is shown.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The supplied code generates a fingerprint from the machine's MAC address rather than querying or analyzing procurement data. Device identification is not disclosed in the declared purpose, and the chunk does not establish that it is necessary to support the advertised functionality. No transmission or persistence of the fingerprint is shown.

Content

No source excerpt is available for this finding.

The query workflow sends user-supplied research targets to an external service without a user-facing privacy disclosure.

Medium
Category
Output Handling
Confidence
91% confidence
Finding

The company-analysis workflow instructs the agent to submit company names and competitive-research targets to the hosted Zhiliao API. Although the API endpoints are documented and device collection during registration requires consent, the skill does not warn users that their query content will be transmitted externally; research targets may reveal confidential business interests.

Content

No source excerpt is available for this finding.

Authentication-error handling directs automatic registration even when an existing key is configured, contradicting the documented registration boundary.

Medium
Category
Excessive Agency
Confidence
94% confidence
Finding

The authentication-error table directs registration even for an existing invalid, expired, or disabled key, contradicting the rule that configured credentials prohibit automatic registration. An agent following that table could initiate unintended device collection or account creation and potentially replace the locally configured account, although the separate registration gate and consent requirement mitigate this risk.

Content

No source excerpt is available for this finding.

The registration workflow collects and transmits a persistent hardware-derived identifier, a capability outside the manifest's tender-data analysis scope.

Medium
Category
Excessive Agency
Confidence
90% confidence
Finding

L80–L83 read a network interface's MAC address and hash it; L109–L125 instruct sending that hash and other device characteristics to an account-registration endpoint. Although the workflow requires consent, hardware-based trial-account deduplication is a separate capability from the advertised tender searches and market analytics, and hashing the MAC still provides a stable device identifier.

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**