Security audit

Reviewer Agent

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed review helper that inspects OpenClaw skill files, runs shell syntax checks, writes review output, and mentions a Feishu notification, with no evidence of hidden or destructive behavior.

Install only if you are comfortable with a review agent reading local OpenClaw skill files, running bash syntax checks, writing review reports, and potentially sending a short conclusion through Feishu. Confirm the target project and notification recipient before using it on confidential work.

SkillSpector

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
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

Findings (2)

Context-Inappropriate Capability

Medium

Confidence: 94% confidence
Finding: The skill instructs the agent to send a Feishu notification even though the metadata does not declare any external messaging capability or related permission boundary. This creates a scope mismatch: an agent following the skill may attempt outbound communication or trigger another tool/workflow without explicit authorization, which can lead to unintended data disclosure or unauthorized notification behavior.

Vague Triggers

Medium

Confidence: 88% confidence
Finding: The trigger phrases include very generic terms such as '评审', 'review', '评分', and '质量检查', which can match ordinary conversation and invoke the skill unexpectedly. Over-broad activation increases the chance that the skill runs in unintended contexts, potentially causing unauthorized file inspection, report generation, or other review actions without clear user intent.

Static analysis

No suspicious patterns detected.