Back to skill

Security audit

商机雷达-比招标更早发现机会

Security checks for vulnerabilities and agentic risk

Overview

The skill’s opportunity-search purpose is coherent, but it also performs vendor account setup, stores an API key locally, sends a device-derived identifier, and can consume paid credits.

Install only if you are comfortable with a Chinese vendor API workflow that can consume account credits, write reports to your home directory, store an API key in ~/.zlbx/config.json, and, if no key is configured, ask to auto-register using platform, CPU architecture, and a hashed MAC-derived device value. Prefer providing your own ZLBX_API_KEY and decline auto-registration if you do not want device-derived data sent to the vendor.

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 (29)

Vague Triggers

High
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill description says it 'must' be used even when the user does not explicitly ask for this skill, including loosely related lead-generation or project-discovery requests. This over-broad activation scope can hijack routing decisions, trigger unexpected external API calls, consume paid credits, and cause unnecessary data disclosure to the vendor API without sufficiently specific user intent.

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,对话清单输出后默认再用 `scripts/render_report.py` 生成一份可分享的 HTML 版商机清单并告知保存路径(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,对话清单输出后默认再用 `scripts/render_report.py` 生成一份可分享的 HTML 版商机清单并告知保存路径(详见 report-template.md「HTML 报告导出」)。

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 · SKILL.md (reported line 171)May include surrounding context.

  • 用户想持续跟踪 → 介绍订阅模式(见上方「订阅模式」)
  • 以上都不贴切 → 询问是否要对某条高价值商机深挖,或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: zhiliao-opportunity-radar/1.0.6

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 · SKILL.md (reported line 171)May include surrounding context.

  • 用户想持续跟踪 → 介绍订阅模式(见上方「订阅模式」)
  • 以上都不贴切 → 询问是否要对某条高价值商机深挖,或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: zhiliao-opportunity-radar/1.0.6

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/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: opportunity-radar/1.0.3

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/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: opportunity-radar/1.0.3

Hidden Instructions

High
Category
Prompt Injection
Confidence
89% confidence
Finding

This duplicate hidden-instruction match is still a valid concern: the markdown embeds agent directives in comments rather than transparent user-visible configuration. Such covert prompt steering increases supply-chain and prompt-injection risk because the model is being guided by content the user may never see.

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: opportunity-radar/1.0.3

Hidden Instructions

High
Category
Prompt Injection
Confidence
89% confidence
Finding

This duplicate hidden-instruction match is still a valid concern: the markdown embeds agent directives in comments rather than transparent user-visible configuration. Such covert prompt steering increases supply-chain and prompt-injection risk because the model is being guided by content the user may never see.

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: opportunity-radar/1.0.3

Hidden Instructions

High
Category
Prompt Injection
Confidence
94% confidence
Finding

This is effectively the same hidden-instruction issue duplicated by the scanner: concealed prompt content directs the agent to use configured credentials and fixed package/source headers on API requests. Because it affects authenticated network behavior, it creates a meaningful prompt-level control channel over sensitive request metadata.

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: opportunity-radar/1.0.3

Hidden Instructions

High
Category
Prompt Injection
Confidence
94% confidence
Finding

This is effectively the same hidden-instruction issue duplicated by the scanner: concealed prompt content directs the agent to use configured credentials and fixed package/source headers on API requests. Because it affects authenticated network behavior, it creates a meaningful prompt-level control channel over sensitive request metadata.

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: opportunity-radar/1.0.3

Hidden Instructions

High
Category
Prompt Injection
Confidence
88% confidence
Finding

This duplicate finding points to the same hidden instruction block that mandates credentialed requests with fixed package/source headers. Even if the headers are legitimate for routing or billing, placing them in concealed prompt content is dangerous because it conditions the agent to obey hidden request-shaping instructions that may conflict with higher-level security boundaries or leak provenance information unexpectedly.

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: opportunity-radar/1.0.3

Hidden Instructions

High
Category
Prompt Injection
Confidence
88% confidence
Finding

This duplicate finding points to the same hidden instruction block that mandates credentialed requests with fixed package/source headers. Even if the headers are legitimate for routing or billing, placing them in concealed prompt content is dangerous because it conditions the agent to obey hidden request-shaping instructions that may conflict with higher-level security boundaries or leak provenance information unexpectedly.

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: opportunity-radar/1.0.3

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 64)May include surrounding context.

python
ROUTE_COLOR = {"拟建": "#7c5cbf", "意向": "#0b7a6a", "临期": "#b9770e"}

# 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 = "iVBORw0KGgoAAAANSUhEUgAAASsAAABgCAYAAABFTFSvAAAAAXNSR0IArs4c6QAAAERlWElmTU0AKgAAAAgAAYdpAAQAAAABAAAAGgAAAAAAA6ABAAMAAAABAAEAAKACAAQAAAABAAABK6ADAAQAAAABAAAAYAAAAACurZhBAAA2V0lEQVR4Ae2dB5xU1fXHzzZ6752ldxHpHTtqTKJJjEmMGkvKP/aIRk0UayzYxZJouokaYyzRiKCAoIB0kN47LB0WFnZh9//9vZmHb4eZnZndmWVW3vl8fjOv3fvuO+++3z3nvHvvM/PF14CvAV8DFUADafbM2OusKOMCKywoqgDlDV/ENIpemJ5vGYW5VmjbrChtlaXZYks7usRuumlb+ET+Vl8DvgYqkgYyzdJ6WqWskXa0IhU7UlnTzLgiK4K8jh7litK22FPPT4O03rb8zAk26hc5kVL6230N+BpIbQ1kWpEdsYJ8Hu4jqV3SeEuXlpZBkhaWmf49s/RvW9aRFfbs2D9ZweF/2i23bIo3O/94XwO+Bk6sBtJP7OmTeHZZV0JBgdmRI1mWnt7V0jIesozK7+H6/shGj66UxLP7Wfsa8DWQYA18fcnKq6iAWxggrbS0XpDWc1a34dP2zDMtvIf5y74GfA2krgZODrJy9e+SVuHROpae8ROC8i/bsy/1cHf7/74GfA2krgZOLrJy70NhoWJ0lXENz8Tces6eHHuqu8v/9zXgayA1NXBykpXuhayswkJeMKQPtHR7greGnVLzFvml8jXga0AaOHnJSlfvENbRLEtLG2BphaPtiSfqabMvvgZ8DaSeBk5ustL9EGEVFVW1jIwzLC3rlyzTWcsXXwO+BlJNAz5Z6Y4ohnXkaCPLSP++PfnCkFS7SX55fA34GjjZ3cDjakBatqUduQZ3sOpxu/wNvgZ8DZxQDfiWlav+gDtY3TKz+plVHupu9v99DfgaSA0N+GTlvQ8Bwmpm6YUX2htvaLiOL74GfA2kiAaST1bEr5Miycq3sLAm5e1jG7e3SUq5/Ux9DfgaKJUGkkNWRxgUfUSTHjAJQjqnKGRZ6wpklyQByybyEZpIQeC4jHQy1/HKV9sSIcovLS0NV7ARhcUd9MXXgK+BVNGAJlRJnEBGVbOy7Lz2be1bbdtax3p1LYtnf1tenk3euMneWrHSVu7cbTCNSAGyUU9yCILV9IxMq5KVaUfZdPioBh9DQDomTXxaZNUyM613s2Z2UYe2dlrDhlaD8+w5nG8zt22zd1atthlbmcZKZChyLKsUFtWxtKJTyOYfZc3KT+9rwNdAYjSQZk8//wKk8POyThGTAfF0b9DQHh022M5p3eq40slr27R/vz06a649N28uYSGzVnXr2vltWtvpLVpY29q1rHqlLDsC4ew8dMjm5+yw91avsbk5260mxHRr3952eZdOVp3lUNHxf1m01B6bNdu25uZChmUINzkEmXYIa+1dGPNKppPJCz2fv+5rwNdA+WsgIZZVBpbPKQ0b2avnn2tdsKaO4E6t3L3bpm/ZagcKjjgW1sAmTaxFzZr20JCB1q1BPQwXw0pqZw2rhu8lMBQr6mendLecgwctl2leOkFsynfdvn32xdYcZ3vzmjWsb+NG1rxGDefYJtWr2k2Tpth20pTJwioqghHTGllWVl1uiU9W5V8v/TP6GjhOAwkhq/pVqth9gwY4RLU3P98enTnHnsd62pN32IkrZeLi9Wve1P5w1unWtX59+1mP7scVJNyGLFw6EZGkEKKat32H3Txxkk1du94ho0xcQ+V3W7/e9oNOHe2c7NZ2Q699ds/n05nduJTCeZAM4m01sK4UbPfF14CvgRTQQJnJqjJhpcHNm9t5uHP
...[truncated 27 chars]

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
92% confidence
Finding

The skill explicitly performs local file writes by generating and saving HTML reports under ~/zlbx-opportunity-radar-files/, but it does not declare any tool scope or permissions boundary for that capability. Undeclared write access weakens platform enforcement and user visibility, creating room for unintended filesystem writes or abuse if the skill is invoked in broader contexts.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

The file's natural-language instructions and examples are written as mandatory Chinese interaction behavior, but do not offer the user any language or locale choice. A skill may be region-specific, but the policy requires that language constraints be documented and justified or made optional for the user.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The instructions require that skill_update_notice.message be relayed to the user verbatim and the surrounding guidance is written as a fixed Chinese-language interaction pattern. This imposes a specific language/locale on user-visible communication without indicating that the user can choose another language or that the constraint is region-specific and justified.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The document materially expands the skill from opportunity discovery into account provisioning, credential storage, and recharge/login lifecycle management. That broadening increases the attack surface by authorizing local secret persistence and account actions unrelated to the declared purpose, making abuse or user-confusing side effects more likely.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill instructs the agent to run OS-specific commands to collect device fingerprints and transmit a hashed MAC-derived identifier for auto-registration. Even with hashing and user consent text, this is host-level data collection unrelated to the core business-search function and creates privacy and tracking risk across sessions and devices.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
86% confidence
Finding

This section explicitly instructs making an outbound registration request carrying device-derived data and receiving an API key. External transmission of locally collected host attributes and credential bootstrap data is security-relevant because it exfiltrates environment information to a third party and enables account creation without manual user-managed onboarding.

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
83% confidence
Finding

The curl example reinforces a direct external POST path for transmitting serialized payloads to the vendor service. While the serialization advice is sound, the underlying behavior is still data exfiltration and remote account bootstrap beyond the skill's declared search purpose.

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
94% confidence
Finding

This duplicate session-persistence finding is also valid because the same instruction persists a remotely issued credential locally and marks it for future automated flows. That persistence compounds risk by enabling subsequent auto-login/recharge behavior without renewed user review.

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
94% confidence
Finding

This duplicate session-persistence finding is also valid because the same instruction persists a remotely issued credential locally and marks it for future automated flows. That persistence compounds risk by enabling subsequent auto-login/recharge behavior without renewed user review.

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**

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The file adds balance checks, update acknowledgements, and auto-login recharge link generation, which are account lifecycle operations outside the stated opportunity-radar function. These capabilities normalize silent account-state handling and external navigation, increasing the chance of unintended billing, phishing-like UX, or secret misuse.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

SQP-3 applies to all file types and covers language or locale policy violations in natural-language instructions. The entire template and required phrasing are Chinese-only, and the file gives prescriptive output instructions in Chinese without indicating that users may choose another language or that the skill is intentionally limited to a Chinese-language regional workflow.

Content

No source excerpt is available for this finding.