Back to skill

Security audit

招标采购信息智能查询

Security checks for vulnerabilities and agentic risk

Overview

The search functions are coherent, but optional registration understates the privacy implications of sending a persistent device identifier.

Review the registration privacy wording before installing. Providing your own API key bypasses device collection; optional registration otherwise sends a stable device fingerprint whose retention and deletion terms are unspecified. Searches send your criteria to Zhiliaobiaoxun and may consume paid credits. Protect the saved key and generated login links. The scanner's missing-reference gaps were resolved locally; remote service behavior was not tested.

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 (14)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The supplied code performs local device identification rather than tender or procurement information retrieval. No connection to a necessary supporting search operation is shown, and device fingerprinting is not disclosed in the declared purpose. The chunk does not show transmission or storage of the resulting hash.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The supplied chunk performs local device fingerprinting rather than tender or procurement information retrieval. Reading and hashing a MAC address is not an evident supporting requirement for the declared search capabilities. No transmission, persistence, or procurement-search behavior is shown.

Content

No source excerpt is available for this finding.

The search workflow does not require a user-facing notice that query criteria are sent to a third-party service.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

The skill instructs the agent to submit searches to the externally hosted Zhiliaobiaoxun API, including user-provided keywords, company names, and other search criteria. Although the endpoints are documented, there is no instruction to disclose this transmission to the user; the consent notice at lines 52–55 covers device characteristics for registration only.

Content

No source excerpt is available for this finding.

Automatic registration collects and transmits device fingerprints not justified by procurement search.

Medium
Category
Excessive Agency
Confidence
94% confidence
Finding

L053-L055 instruct the agent to collect platform, architecture, and a MAC-address hash and send them to an account-registration endpoint. The manifest at L003 describes procurement-information retrieval, not device identification; although prior user consent is required, this collection is an additional privacy-sensitive capability rather than an obvious requirement of searching procurement records.

Content

No source excerpt is available for this finding.

Standalone corporate-registry investigations exceed the manifest's procurement-information retrieval scope.

Medium
Category
Excessive Agency
Confidence
91% confidence
Finding

L076-L080 expose queries for corporate registration, shareholders, branches, investments, and registration changes, while L542-L546 additionally expose annual reports, certificates, and credit ratings. These are standalone corporate-information capabilities, not merely ways to filter or retrieve tender announcements; L103 expressly supports shareholder-only requests without querying procurement information.

Content

No source excerpt is available for this finding.

The skill mandates an unrelated package-update notification and acknowledgment workflow.

Medium
Category
Excessive Agency
Confidence
91% confidence
Finding

The manifest describes tender and procurement information retrieval, but these instructions require a first-use account request to check upgrade notices, verbatim delivery of server-provided update messages, and an authenticated POST acknowledging their display. This package-maintenance workflow introduces unsolicited user-facing content and remote acknowledgment state that are not necessary to retrieve procurement information.

Content

No source excerpt is available for this finding.

L393–L395 mandate a package-upgrade notification and acknowledgment workflow unrelated to tender searches.

Medium
Category
Excessive Agency
Confidence
92% confidence
Finding

The instructions require an authenticated account request at the first use in each session to check upgrade notices, relay the returned message, and submit a POST acknowledgment after displaying it. This introduces package-maintenance messaging and server-side acknowledgment state changes beyond the manifest’s stated procurement-information search and retrieval purpose.

Content

No source excerpt is available for this finding.

The skill defines standalone corporate ownership and financial-report queries beyond its declared tender-search purpose.

Medium
Category
Excessive Agency
Confidence
95% confidence
Finding

The manifest describes searching procurement announcements, retrieving tender details, and tracking projects. These instructions additionally enable paid queries for corporate shareholders, investments, registration changes, and annual reports containing financial and contact information; these standalone corporate-intelligence capabilities are not necessary to perform the declared tender-search functions.

Content

No source excerpt is available for this finding.

The skill mandates an account-based upgrade-notification and acknowledgement workflow unrelated to procurement searches.

Medium
Category
Excessive Agency
Confidence
92% confidence
Finding

L334 requires an authenticated account-balance request at the first use in each session specifically to check upgrade notices. L335–L336 then require forwarding the notice and submitting an authenticated POST acknowledgement, introducing account access and server-side notification-state changes beyond the manifest's procurement search and retrieval purpose.

Content

No source excerpt is available for this finding.

The skill collects and transmits a persistent hardware-derived device fingerprint, a capability beyond its stated tender-search purpose.

Medium
Category
Excessive Agency
Confidence
93% confidence
Finding

An unsalted SHA-256 hash of a MAC address remains a persistent, linkable device identifier; hashing does not make it anonymous, and the constrained MAC address space permits candidate matching. The explicit consent gate and manual-key alternative reduce risk, but describing the collected characteristics as having no identifying meaning undermines informed consent for hardware fingerprinting beyond the tender-search function.

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 mandates an unrelated package-update and acknowledgement workflow during procurement searches.

Low
Category
Excessive Agency
Confidence
93% confidence
Finding

The manifest describes tender and procurement information retrieval, but these instructions require an authenticated account request on the first use of every session to check package-update notices. They also require displaying the notice and submitting a POST acknowledgement that changes the server's reminder state, neither of which is necessary to retrieve procurement information or declared in the manifest.

Content

No source excerpt is available for this finding.