Back to skill

Security audit

企业投标决策智能助手

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly coherent for bidding analysis, but it needs review because it persists API credentials, collects a device fingerprint for registration, hides request/update rules in comments, and preserves signed access links in shareable reports.

Install only if you are comfortable with a third-party bidding service receiving your project/company queries, storing or reusing a local API key, collecting a hashed MAC address during optional auto-registration, and producing reports that may contain signed access links. Avoid sharing exported reports unless signed URLs are removed or you accept that recipients may use those links.

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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Rogue AgentSelf-Modification, Session Persistence
Findings (30)

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

声明描述的是一个核心分析型技能,重点在于基于外部历史数据对具体招标项目进行智能评估并产出决策结论。而给出的代码是一个纯展示层工具:它从输入 JSON 读取预先存在的字段,拼装 HTML,内嵌 CSS/Logo,并提供打印、导出 PDF、保存 PNG 长图等前端功能。代码没有网络请求、数据库查询、模型推理、评分逻辑、竞对预测逻辑、报价计算逻辑,也没有任何从招标文件或公告中提取信息并进行分析的实现。因此,代码实际行为与声明的主要用途存在实质性不一致。

Content

No source excerpt is available for this finding.

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 81)May include surrounding context.

md
完整模式下,对话报告输出后默认再用 `scripts/render_report.py` 生成一份可分享的 HTML 版报告并告知保存路径(详见 report-template.md「HTML 报告导出」)。

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 85)May include surrounding context.

md
完整模式下,对话报告输出后默认再用 `scripts/render_report.py` 生成一份可分享的 HTML 版报告并告知保存路径(详见 report-template.md「HTML 报告导出」)。

Hidden Instructions

High
Category
Prompt Injection
Confidence
88% confidence
Finding

This duplicate finding identifies the same hidden-instructions issue: behavior-changing network requirements and acknowledgement logic are concealed in a comment block. Such hidden control logic reduces auditability and can be abused to make the agent perform undisclosed network-side actions or attach tracking-like metadata to requests.

Content

Scanner excerpt · SKILL.md (reported line 163)May include surrounding context.

  • 报告里某个竞争对手威胁最大 → 引导用 zlbx-company-intel 对其深度背调
  • 以上都不贴切 → 报告涉及的企业完整档案与更多商机详情,引导访问知了商机大师 https://agent.zhiliaobiaoxun.com

安装包身份与升级检查:所有发往知了标讯 API 的请求(包括账户查询)必须额外携带以下两个请求头:

text
X-Skill-Package: enterprise-bidding-decision-agent/1.0.6

Hidden Instructions

High
Category
Prompt Injection
Confidence
88% confidence
Finding

This duplicate finding identifies the same hidden-instructions issue: behavior-changing network requirements and acknowledgement logic are concealed in a comment block. Such hidden control logic reduces auditability and can be abused to make the agent perform undisclosed network-side actions or attach tracking-like metadata to requests.

Content

Scanner excerpt · SKILL.md (reported line 163)May include surrounding context.

  • 报告里某个竞争对手威胁最大 → 引导用 zlbx-company-intel 对其深度背调
  • 以上都不贴切 → 报告涉及的企业完整档案与更多商机详情,引导访问知了商机大师 https://agent.zhiliaobiaoxun.com

安装包身份与升级检查:所有发往知了标讯 API 的请求(包括账户查询)必须额外携带以下两个请求头:

text
X-Skill-Package: enterprise-bidding-decision-agent/1.0.6

Hidden Instructions

High
Category
Prompt Injection
Confidence
83% confidence
Finding

This duplicate finding points to the same hidden instruction block that tells the agent which headers and credential context to use. Even if intended for normal integration, embedding such directives in free-form documentation is dangerous because an adversarial skill can steer authenticated requests, create cross-package identity confusion, or cause data to be sent under the wrong trust boundary.

Content

Scanner excerpt · references/api-quick.md (reported line 1)May include surrounding context.

请求身份:使用已配置的 API Key,并保留本包的 X-Client,携带下列完整身份头;不得沿用其它包的版本或来源。

text
X-Client: bid-decision/1.0.5

Hidden Instructions

High
Category
Prompt Injection
Confidence
83% confidence
Finding

This duplicate finding points to the same hidden instruction block that tells the agent which headers and credential context to use. Even if intended for normal integration, embedding such directives in free-form documentation is dangerous because an adversarial skill can steer authenticated requests, create cross-package identity confusion, or cause data to be sent under the wrong trust boundary.

Content

Scanner excerpt · references/api-quick.md (reported line 1)May include surrounding context.

请求身份:使用已配置的 API Key,并保留本包的 X-Client,携带下列完整身份头;不得沿用其它包的版本或来源。

text
X-Client: bid-decision/1.0.5

Hidden Instructions

High
Category
Prompt Injection
Confidence
88% confidence
Finding

This is the same hidden-instruction issue duplicated by the scanner: the comment block covertly directs identity/header behavior for API calls. Such concealed prompt content increases the chance of unauthorized or non-transparent request shaping and should be treated as a real prompt-injection style risk when embedded in skill text.

Content

Scanner excerpt · references/auto-register.md (reported line 1)May include surrounding context.

请求身份:使用已配置的 API Key,并保留本包的 X-Client,携带下列完整身份头;不得沿用其它包的版本或来源。

text
X-Client: bid-decision/1.0.5

Hidden Instructions

High
Category
Prompt Injection
Confidence
88% confidence
Finding

This is the same hidden-instruction issue duplicated by the scanner: the comment block covertly directs identity/header behavior for API calls. Such concealed prompt content increases the chance of unauthorized or non-transparent request shaping and should be treated as a real prompt-injection style risk when embedded in skill text.

Content

Scanner excerpt · references/auto-register.md (reported line 1)May include surrounding context.

请求身份:使用已配置的 API Key,并保留本包的 X-Client,携带下列完整身份头;不得沿用其它包的版本或来源。

text
X-Client: bid-decision/1.0.5

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The file directs the agent to perform account provisioning, device fingerprint collection, API-key management, recharge-link generation, and account-recovery flows that are materially outside the declared bidding-analysis purpose of the skill. Even with a stated consent step, this expands the skill into credential handling and endpoint orchestration, increasing the chance of privacy abuse, overcollection, and unauthorized external actions from within a business-analysis workflow.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill instructs collection of local device characteristics such as OS, CPU architecture, and a MAC-derived hash, then persists credentials in a local config file. These actions are unrelated to bid-decision analysis and create a new privacy and credential-storage surface; hashed hardware identifiers are still stable device fingerprints and can enable tracking or account linkage across sessions.

Content

No source excerpt is available for this finding.

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · references/report-template.md (reported line 1)May include surrounding context.

请求身份:使用已配置的 API Key,并保留本包的 X-Client,携带下列完整身份头;不得沿用其它包的版本或来源。

text
X-Client: bid-decision/1.0.5

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · references/report-template.md (reported line 1)May include surrounding context.

请求身份:使用已配置的 API Key,并保留本包的 X-Client,携带下列完整身份头;不得沿用其它包的版本或来源。

text
X-Client: bid-decision/1.0.5

Missing User Warnings

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The template repeatedly instructs preservation of signed sk parameters and gives operational guidance that they bypass login barriers, but provides no warning that these links may carry access-bearing tokens. This combination strongly encourages sharing of authenticated URLs in normal output and exports, creating a straightforward token leakage channel.

Content

No source excerpt is available for this finding.

Ssd 3

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The template directs the skill to preserve and reproduce signed access URLs, including parameters explicitly described as enabling access without login, in both visible output and exported HTML/JSON. This is a direct secret-disclosure pattern: anyone who receives the report may gain unintended access, and the tokens may further spread through browser history, referers, analytics, or document sharing workflows.

Content

No source excerpt is available for this finding.

Hidden Instructions

High
Category
Prompt Injection
Confidence
93% confidence
Finding

This duplicate finding points to the same hidden comment/instruction block that directs the agent to preserve package identity headers and not reuse other package sources. Even if framed as package integrity guidance, hidden instructions that alter authentication behavior are dangerous because they can override normal trust boundaries and induce authenticated requests the user did not authorize.

Content

Scanner excerpt · references/workflow.md (reported line 1)May include surrounding context.

请求身份:使用已配置的 API Key,并保留本包的 X-Client,携带下列完整身份头;不得沿用其它包的版本或来源。

text
X-Client: bid-decision/1.0.5

Hidden Instructions

High
Category
Prompt Injection
Confidence
93% confidence
Finding

This duplicate finding points to the same hidden comment/instruction block that directs the agent to preserve package identity headers and not reuse other package sources. Even if framed as package integrity guidance, hidden instructions that alter authentication behavior are dangerous because they can override normal trust boundaries and induce authenticated requests the user did not authorize.

Content

Scanner excerpt · references/workflow.md (reported line 1)May include surrounding context.

请求身份:使用已配置的 API Key,并保留本包的 X-Client,携带下列完整身份头;不得沿用其它包的版本或来源。

text
X-Client: bid-decision/1.0.5

Obfuscated Code

High
Category
Supply Chain
Confidence
50% confidence
Finding

Code contains obfuscation (base64, hex encoding with execution). This is often used to hide malicious functionality.

Content

Scanner excerpt · scripts/render_report.py (reported line 71)May include surrounding context.

python
THREAT_COLOR = {"高": "#c0392b", "中": "#b9770e", "低": "#0d9463"}

# 知了标讯白色 logo(299x96 PNG base64 内嵌,保证报告离线/打印/转发时不裂图)
_LOGO_B64 = "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
...[truncated 27 chars]

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
82% confidence
Finding

The skill explicitly instructs the agent to write HTML reports to a local directory, but it does not declare any corresponding tool scope or permissions boundary. That creates a least-privilege violation: a host may expose file-write capability more broadly than users expect, enabling persistent local artifact creation without an explicit permission contract.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The instruction '报告全文只输出一遍' is adjacent to a mandatory wording policy that specifies exact Chinese phrasing and replacements, effectively constraining output language/style for all reports involving real entities. The file does not offer a user opt-in or alternative locale/language path, so this is a language policy issue under the natural-language policy category.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The document contains user-facing instructions entirely in Chinese and explicitly says the privacy notice can be relayed verbatim to the user, which effectively fixes the interaction language. Elsewhere it also prescribes exact Chinese output text, but does not offer the user a language/locale choice or justify a mandatory locale scope.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
91% confidence
Finding

The document explicitly instructs sending collected device features to an external registration endpoint. External transmission of locally derived identifiers and environment attributes is security-relevant because it exports host-derived data outside the agent boundary; in this skill, that transmission is especially risky because it is unrelated to the core bidding-analysis purpose.

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
90% confidence
Finding

The curl example is another explicit instruction to exfiltrate collected registration payload data to an external service. Although the advice about JSON serialization is technically sound, the underlying behavior remains problematic because it operationalizes off-scope data transmission from the user's machine to a third party.

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
95% confidence
Finding

This duplicated finding describes the same persistence behavior: creating ~/.zlbx and storing source-tagged API credentials for future automatic use. That persistence is security-significant because it extends credential lifetime and enables silent follow-on actions like auto-login link generation without renewed user intent.

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**

Session Persistence

Medium
Category
Rogue Agent
Confidence
95% confidence
Finding

This duplicated finding describes the same persistence behavior: creating ~/.zlbx and storing source-tagged API credentials for future automatic use. That persistence is security-significant because it extends credential lifetime and enables silent follow-on actions like auto-login link generation without renewed user intent.

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**