Security audit

免费版

Security checks for vulnerabilities and agentic risk

Overview

The skill is a coherent contract-review prompt, but it handles sensitive legal documents while understating that content may go to the user's configured external AI provider.

Install only if you are comfortable sending contract contents to the AI provider configured in OpenClaw. Avoid uploading highly confidential, privileged, or regulated documents unless you have checked that provider's retention and privacy terms, and explicitly ask the agent before processing sensitive files.

A.I.G

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

SkillSpector

Vulnerability Patterns

Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (4)

Intent-Code Divergence

Medium

Confidence: 89% confidence
Finding: The skill tells users their files are not stored with third parties, while also stating documents are sent to whatever external AI provider the user configured. That can mislead users about where sensitive contract data goes and what third-party retention or training policies may apply, especially for confidential legal documents.

Vague Triggers

Medium

Confidence: 95% confidence
Finding: The trigger list includes broad phrases such as '看看这个合同' and activation on any uploaded contract-like document, which can cause the skill to run when the user did not explicitly intend a legal review. In a skill that processes sensitive legal documents, unintended activation increases the chance of unnecessary disclosure of confidential content to the model pipeline or external providers.

Natural-Language Policy Violations

Medium

Confidence: 98% confidence
Finding: The instruction "For Chinese contracts (中文合同), output the entire analysis in Chinese" imposes a language choice based solely on document language, not user preference. This is a natural-language locale policy issue because it does not offer the user an option to receive the analysis in another language.

Vague Triggers

Medium

Confidence: 90% confidence
Finding: The listed trigger phrases include very general everyday requests such as “帮我对比一下这两份合同的区别” and “这份合同能签吗，帮我简单看一下,” but the document does not define what exactly causes the skill to activate or what requests should not trigger it. Without explicit scope constraints or negative examples, these phrases may overlap with ordinary chat about documents and create ambiguous invocation behavior.

Static analysis

No suspicious patterns detected.