Back to skill

Security audit

物业纠纷全流程维权

Security checks for vulnerabilities and agentic risk

Overview

This skill is coherent for China-focused property-dispute help, but it creates sensitive local HTML reports without clear safety controls and includes legally risky public-exposure guidance.

Review before installing. Use this only for China property-management disputes, avoid entering unnecessary personal details, inspect generated complaint text carefully, and do not publish accusations or identifying information without verifying facts and considering legal advice. Treat generated HTML reports as sensitive files and avoid opening reports built from untrusted input unless the template is hardened.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (1)

T09 · Insecure Skill Coding Practices

Warning
Location
assets/report_template.html:143
Finding

Unescaped User-Controlled Content in Generated HTML Report

Content
View full analysis

📜 二、法律依据

{{LEGAL_CARDS}}

🪜 三、维权升级流程

{{TIMELINE_ITEMS}}

📝 四、投诉文件

{{COMPLAINT_DOCUMENTS}}

💬 五、谈判脚本

{{NEGOTIATION_SCRIPTS}}

💰 六、公共收益与税务审计

{{TAX_AUDIT}}

⚠️ 以上内容为法律信息参考,不构成正式法律意见。建议在采取重大法律行动前咨询专业律师。

建议立即采取以下行动:

{{ACTION_RECOMMENDATIONS}}
``` The same issue also affects `{{REPORT_TITLE}}` in `assets/report_template.html:6` and `assets/report_template.html:112`. ### Technical Analysis The skill directs the agent to replace every template placeholder with generated content and then save the result as an HTML file. Several placeholders can include information derived from a user's dispute description, complaint documents, negotiation scripts, and recommendations. No instruction requires HTML escaping, contextual output encoding, or sanitization before substitution. Consequently, attacker-controlled strings containing HTML can be interpreted as markup rather than displayed as text. If active elements or attributes survive generation, the resulting report may contain script-capable event handlers, deceptive forms ...[truncated 1983 chars]
Remediation
View remediation
`, `"`, and `'`. - Use attribute-specific encoding if a value is ever inserted into an HTML attribute. 2. **Sanitize intentionally generated rich HTML** - If placeholders such as `LEGAL_CARDS` must contain markup, generate that markup from trusted components rather than concatenating arbitrary strings. - Pass rich content through a strict allowlist sanitizer. - Permit only required structural elements and safe attributes. - Remove `
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (5)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The trigger list includes broad, everyday phrases such as '漏水', '物业问题', and 'property dispute', which can cause the skill to activate in contexts where the user did not intend to invoke a formal legal-rights workflow. In this skill, unintended activation is more concerning because it can lead to collection of sensitive dispute details and generation of authoritative-looking legal/escalation guidance, increasing the chance of overreach or user confusion.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill instructs the agent to save an HTML report into the current working directory and present the file, but it does not require prior user notice or consent for local file creation. In a dispute-assistance context, the report may contain sensitive legal, financial, and complaint-related information, so silent persistence increases privacy risk and can leave residual data on the host environment.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The HTML document is explicitly marked with lang="zh-CN", and all visible user-facing content in the template is written in Chinese. This imposes a specific language/locale on users without any opt-in or indication that the skill is intentionally limited to a China-specific audience, which matches the language/locale policy violation criteria.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The document explicitly recommends media exposure, naming specific companies and communities, and creating public pressure, but it does not warn about privacy, defamation, doxxing, evidentiary uncertainty, or the irreversible nature of online publication. In a dispute-resolution skill, this can push users toward legally risky public accusations before facts are verified, increasing harm to both the target and the user.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
81% confidence
Finding

The entire skill content is presented only in Chinese and references Chinese civic channels, with no statement that the content is region-specific or that language/locale is intentionally constrained. Under the policy rule, forcing a specific language without opt-in or documented justification is a natural-language policy concern.

Content

No source excerpt is available for this finding.