Back to skill

Security audit

竞对情报分析-投标对手底细摸排

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a disclosed competitor-intelligence integration, but it also handles API credentials, device fingerprinting, signed access links, and hidden request instructions in ways users should review before installing.

Install only if you are comfortable with this skill contacting zhiliaobiaoxun.com using your API key, checking account status, optionally creating and storing an API key after consent, collecting a hashed MAC-derived identifier for trial registration, and saving reports that may contain signed access links. Avoid sharing generated HTML reports unless you are comfortable exposing the embedded sk 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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Findings (36)

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The skill promises substantive competitor-intelligence analysis from procurement data, but the detected behavior indicates mainly HTML rendering/export rather than the claimed data retrieval and analytical workflow. This mismatch is dangerous because users may trust outputs as evidence-based intelligence when the implementation may simply format provided input, enabling misleading business decisions and concealment of missing analysis.

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 89)May include surrounding context.

md
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 93)May include surrounding context.

md
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
90% confidence
Finding

This duplicate hidden-instructions finding identifies the same concealed package-identity block, which influences every API call and post-display acknowledgement flow. In a skill that handles account checks, API keys, and external requests, concealed metadata requirements increase the risk of undisclosed tracking and make security review materially harder.

Content

Scanner excerpt · SKILL.md (reported line 185)May include surrounding context.

  • 用户想持续跟踪该公司 → 介绍竞对动态周报(见上方「竞对动态周报」)
  • 以上都不贴切 → 询问是否要对某个竞对做双公司对比,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: competitor-intel-analysis/1.0.6

Hidden Instructions

High
Category
Prompt Injection
Confidence
90% confidence
Finding

This duplicate hidden-instructions finding identifies the same concealed package-identity block, which influences every API call and post-display acknowledgement flow. In a skill that handles account checks, API keys, and external requests, concealed metadata requirements increase the risk of undisclosed tracking and make security review materially harder.

Content

Scanner excerpt · SKILL.md (reported line 185)May include surrounding context.

  • 用户想持续跟踪该公司 → 介绍竞对动态周报(见上方「竞对动态周报」)
  • 以上都不贴切 → 询问是否要对某个竞对做双公司对比,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: competitor-intel-analysis/1.0.6

Hidden Instructions

High
Category
Prompt Injection
Confidence
85% confidence
Finding

This duplicate finding identifies the same hidden instruction pattern: HTML comments carry request-identity directives that are not plainly visible in normal rendered content. In an agent skill, concealed instructions materially increase prompt-injection risk because the model may obey unseen directions about credential use, outbound requests, or package identity without user awareness.

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: company-intel/1.0.2

Hidden Instructions

High
Category
Prompt Injection
Confidence
85% confidence
Finding

This duplicate finding identifies the same hidden instruction pattern: HTML comments carry request-identity directives that are not plainly visible in normal rendered content. In an agent skill, concealed instructions materially increase prompt-injection risk because the model may obey unseen directions about credential use, outbound requests, or package identity without user awareness.

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: company-intel/1.0.2

Hidden Instructions

High
Category
Prompt Injection
Confidence
90% confidence
Finding

This duplicate finding points to the same hidden request-identity directive, which is still a true issue because it embeds concealed behavioral instructions in the skill content. Even if the headers are operational rather than overtly malicious, hidden instructions reduce transparency and can be used to manipulate how the agent authenticates and identifies itself to external services.

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: company-intel/1.0.2

Hidden Instructions

High
Category
Prompt Injection
Confidence
90% confidence
Finding

This duplicate finding points to the same hidden request-identity directive, which is still a true issue because it embeds concealed behavioral instructions in the skill content. Even if the headers are operational rather than overtly malicious, hidden instructions reduce transparency and can be used to manipulate how the agent authenticates and identifies itself to external services.

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: company-intel/1.0.2

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The file adds an account-registration, device-fingerprinting, credential management, and recharge workflow to a skill whose declared purpose is competitor-intelligence analysis. This is a scope-expansion vulnerability because it instructs the agent to access local environment state, collect host-derived identifiers, create accounts, persist secrets, and manage billing-related flows that are unrelated to the user’s analytical task. In an LLM-agent setting, this broadens the attack surface substantially and normalizes privileged actions the user may not expect.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The skill instructs collection of device characteristics including OS, CPU architecture, and a hashed MAC-derived identifier for free-trial registration. Even if framed as privacy-minimized, this is still host fingerprinting unrelated to competitor analysis and causes external transmission of persistent device-linked data from the user’s machine. In agent environments, local hardware/network identifiers are sensitive because they can enable tracking, correlation across sessions, and unexpected exfiltration beyond the skill’s stated purpose.

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: company-intel/1.0.2

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: company-intel/1.0.2

Ssd 3

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

Requiring the agent to retain and display raw URLs with sk parameters directly to users exposes access-bearing tokens in clear text. In the context of a reporting skill that encourages sharing reports and links, this is especially dangerous because the URLs can escape the original session and grant unintended access.

Content

No source excerpt is available for this finding.

Ssd 3

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The export flow instructs the agent to write reports containing access-bearing URLs into JSON/HTML files and then reveal the absolute filesystem path to the user. This compounds token leakage with persistent local storage and path disclosure, making sensitive links easier to copy, exfiltrate, or discover later.

Content

No source excerpt is available for this finding.

Hidden Instructions

High
Category
Prompt Injection
Confidence
95% confidence
Finding

This duplicated hidden-instruction finding points to the same issue: the skill content contains covert instructions that direct the agent to use existing credentials and attach package/source headers on all requests. Even if intended for legitimate integration, placing these directives inside the skill file creates an instruction-channel vulnerability where untrusted content can influence authentication behavior and external communications. Given the skill’s broad workflow and repeated mandate to preserve package identity, exploitation could lead to unauthorized API usage, confused-deputy behavior, or credential misuse.

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: company-intel/1.0.2

Hidden Instructions

High
Category
Prompt Injection
Confidence
95% confidence
Finding

This duplicated hidden-instruction finding points to the same issue: the skill content contains covert instructions that direct the agent to use existing credentials and attach package/source headers on all requests. Even if intended for legitimate integration, placing these directives inside the skill file creates an instruction-channel vulnerability where untrusted content can influence authentication behavior and external communications. Given the skill’s broad workflow and repeated mandate to preserve package identity, exploitation could lead to unauthorized API usage, confused-deputy behavior, or credential misuse.

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: company-intel/1.0.2

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 93)May include surrounding context.

python
# 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/n05nduJTCeZAM4m01sK4UbPfF14CvgRTQQJnJqjJhpcHNm9t5uHP78wvsqTlz7SHIwnHFIBMFwasSixrOMY2qVS92yUcKi2zFnt02f/tOW757j23PO2hZxL
...[truncated 27 chars]

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
88% confidence
Finding

The skill explicitly states it will write HTML reports to a local directory, but it does not declare a tool scope or permissions boundary for file writing. Missing scope declarations weaken least-privilege guarantees and can allow broader-than-expected local file creation if the host runtime infers capabilities implicitly.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The invocation rules are overly broad, instructing use whenever a user wants to understand a company or competitor in bidding contexts, even if they did not explicitly request this skill. Overbroad routing can cause unnecessary external data sharing, unintended account usage, and surprise automation such as report generation or registration prompts for ordinary company-analysis requests.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The file requires that certain content be relayed '原样告知用户一次,' and the overall user-facing examples and mandated phrasing are exclusively in Chinese, without any opt-in or language-selection mechanism. This can violate a language/locale policy when the user has not chosen Chinese as their preferred language.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The skill explicitly instructs the agent to send a configured API key and package identity headers to an external service, but the user-facing content shown here does not require disclosure or consent before outbound credentialed requests. That creates a real transparency and consent problem: users may trigger third-party network access and account-scoped actions without understanding that their query and the configured credentials will be used externally.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The package requires an automatic account-balance check on first use, which causes an outbound request before performing the user's primary task, yet the instructions contain no requirement to inform the user beforehand. This is risky because it silently contacts a third-party service and consumes configured credentials even when the user may not expect any external access at that stage.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
92% confidence
Finding

This section explicitly directs the agent to serialize collected device data and send it in a POST request to an external service. The transmission itself is the risky behavior: it operationalizes exfiltration of local system-derived identifiers to a third party, and the surrounding instructions are detailed enough to make the flow actionable by an agent. In the context of a competitor-analysis skill, this outbound transfer is unjustified and therefore dangerous.

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 alternative instruction to use curl -d reinforces and enables the same external data transmission path using command-line tooling. Providing multiple implementation paths lowers friction for execution by an agent and increases the likelihood that local device-derived data will be sent off-host. Because the skill’s business purpose is unrelated to registration or telemetry, this remains an unjustified exfiltration channel.

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