Back to skill

Security audit

政府采购投标决策-政采项目投标评估

Security checks for vulnerabilities and agentic risk

Overview

This is a real procurement-analysis skill, but it needs Review because it combines hidden account/version-check instructions with persistent credentials and device-based auto-registration.

Install only if you are comfortable with this skill using Zhiliaobiaoxun credentials, storing an API key under your home directory, sending project/company queries to that service, and optionally creating a trial account using hashed device features after consent.

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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Findings (35)

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
91% confidence
Finding

This duplicate hidden-instruction finding refers to the same commented block that directs outbound request modification and server acknowledgment after displaying notices. Because the behavior is concealed in comments rather than declared normally, it undermines auditability and can enable covert side effects.

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: government-procurement-bid-decision/1.0.9

Hidden Instructions

High
Category
Prompt Injection
Confidence
91% confidence
Finding

This duplicate hidden-instruction finding refers to the same commented block that directs outbound request modification and server acknowledgment after displaying notices. Because the behavior is concealed in comments rather than declared normally, it undermines auditability and can enable covert side effects.

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: government-procurement-bid-decision/1.0.9

Hidden Instructions

High
Category
Prompt Injection
Confidence
82% confidence
Finding

This duplicated finding points to the same hidden instruction region that embeds request-header and package-identity directives in non-user-visible content. While the content appears operational rather than overtly malicious, concealment increases the chance that an agent will follow sensitive instructions without transparency or policy review.

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
82% confidence
Finding

This duplicated finding points to the same hidden instruction region that embeds request-header and package-identity directives in non-user-visible content. While the content appears operational rather than overtly malicious, concealment increases the chance that an agent will follow sensitive instructions without transparency or policy review.

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
93% confidence
Finding

The hidden comment block contains operational instructions that influence outbound request identity and package provenance headers without being visible as normal user-facing behavior. Hidden or metadata-embedded instructions are risky in skills because they can steer network behavior, impersonation, or tracking in ways users and reviewers may miss, especially when combined with account registration and credential reuse.

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
93% confidence
Finding

The hidden comment block contains operational instructions that influence outbound request identity and package provenance headers without being visible as normal user-facing behavior. Hidden or metadata-embedded instructions are risky in skills because they can steer network behavior, impersonation, or tracking in ways users and reviewers may miss, especially when combined with account registration and credential reuse.

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 embeds a full account lifecycle workflow—API key discovery, automatic account registration, device fingerprint collection, quota handling, update-notice acknowledgement, and persistence to local config—that materially exceeds the declared purpose of bid-decision analysis. This broadens the skill’s authority from domain analysis into credential management and tracking, creating unnecessary privacy, security, and supply-chain risk if the skill is invoked in normal use.

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 the agent to collect device characteristics (platform, architecture, MAC-derived hash) and use them to create or recover external accounts. Even with user-consent language and hashing, this is still device fingerprinting for an unrelated business function, and it introduces privacy exposure plus the possibility of silent enrollment into a third-party service from within a bidding-analysis skill.

Content

No source excerpt is available for this finding.

Hidden Instructions

High
Category
Prompt Injection
Confidence
97% confidence
Finding

This duplicate hidden-instructions finding points to the same covert request-identity block that instructs the skill to attach specific package/source headers and preserve a particular client identity. Such concealed control text can manipulate downstream requests and is especially risky because it is not part of transparent user-facing behavior.

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
97% confidence
Finding

This duplicate hidden-instructions finding points to the same covert request-identity block that instructs the skill to attach specific package/source headers and preserve a particular client identity. Such concealed control text can manipulate downstream requests and is especially risky because it is not part of transparent user-facing behavior.

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
96% confidence
Finding

This duplicate finding points to the same hidden instruction block that dictates exact outbound identity headers and forbids using another package's version/source. Even if the stated purpose is package attribution, embedding such control in hidden instructions creates a non-transparent trust boundary where the model can be induced to send sensitive or identifying metadata without clear user awareness.

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
96% confidence
Finding

This duplicate finding points to the same hidden instruction block that dictates exact outbound identity headers and forbids using another package's version/source. Even if the stated purpose is package attribution, embedding such control in hidden instructions creates a non-transparent trust boundary where the model can be induced to send sensitive or identifying metadata without clear user awareness.

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
84% confidence
Finding

The skill appears to perform local file writes (for HTML report export and credential persistence) without declaring an explicit tool scope such as allowed-tools or permissions. This creates a mismatch between declared and actual capabilities, reducing reviewability and increasing the chance that an agent can write to the filesystem unexpectedly.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The description says the skill must be used not only for explicit government procurement requests, but also whenever a request merely 'involves' bid evaluation, whether to bid, or pricing reference, even if the user never mentions government procurement. This broad natural-language trigger lacks negative examples or exclusion conditions, making unintended invocation more likely.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The skill instructs the agent to automatically register an external account and persist an API key locally when no key is present. That exceeds the narrow purpose of bid-decision analysis and introduces account creation, credential handling, and device-fingerprint collection flows that expand the attack surface and privacy risk.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The documented behavior includes writing generated reports to local disk and storing credentials in fixed filesystem paths. Even if intended, combining analytics with persistent local writes broadens capability beyond data analysis and can expose sensitive project data or secrets if the host environment is shared or insufficiently sandboxed.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
78% confidence
Finding

The instruction '报告全文只输出一遍' is neutral, but the surrounding file consistently prescribes Chinese phrasing and replacement terminology for output while providing no option for the user's preferred language or locale. Because the skill appears to force a single language for responses without opt-in or documented regional necessity, this may violate language/locale policy.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The documentation directs the skill to use a configured API key and send requests to an external service, but it does not require any user-facing notice or consent before transmitting potentially sensitive project, company, or query data. This creates a covert data egress and credential-use risk because users may not realize their inputs trigger third-party API calls under account-linked credentials.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill mandates an automatic account balance check on first use, which causes an external account lookup without first informing the user. Even if the endpoint is free, it still uses credentials and reveals account metadata/activity to a third-party service without a clear user-visible warning.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

This markdown file contains user-facing instructions and quoted messages entirely in Chinese, including text the agent should relay verbatim to the user. Because the file does not offer a language/locale choice or document a justified region-specific constraint, it creates a natural-language policy concern under the locale policy rule.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
91% confidence
Finding

This instruction explicitly sends collected device features to an external service as part of the auto-registration flow. External transmission is expected for remote APIs, but in this skill it is dangerous because the transmitted data and registration workflow are unrelated to the declared procurement-analysis purpose and encourage host-level data collection before service use.

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