ClawHub
Back to skill

Security audit

巡店管理全流程助手

Security checks for vulnerabilities and agentic risk

Overview

This store-inspection skill is a coherent local reporting tool that creates inspection JSON files and HTML reports, with a disclosed CDN dependency but no evidence of exfiltration or destructive behavior.

Install this if you are comfortable with a local store-inspection tool creating JSON data and HTML reports in its own folders. Open generated HTML reports with the understanding that they load Chart.js from a third-party CDN; avoid placing sensitive business data in reports if that network dependency is unacceptable.

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 (4)

Context-Inappropriate Capability

Medium
Confidence
89% confidence
Finding
The HTML report loads Chart.js from a third-party CDN at report view time, which creates an unnecessary external dependency for an otherwise local inspection workflow. If the CDN is unavailable, blocked, or serves compromised content, anyone opening the generated report could execute untrusted JavaScript in their browser and expose report contents or browser context.

Vague Triggers

Medium
Confidence
88% confidence
Finding
The trigger list is very broad and contains generic retail-inspection phrases without clear scope boundaries, exclusions, or activation constraints. In an agent ecosystem, this can cause the skill to activate in unintended contexts, leading to overreach, workflow hijacking, or inappropriate handling of adjacent tasks that the skill was not meant to perform.

Vague Triggers

Medium
Confidence
83% confidence
Finding
The trigger list contains broad phrases such as planning, analysis, evaluation, and data-related terms that can match many ordinary conversations outside store inspection. Over-broad activation increases the chance of unintended skill invocation, which is more dangerous here because the skill can write files and generate reports, potentially causing side effects from benign user requests.

Missing User Warnings

Medium
Confidence
91% confidence
Finding
The skill states that it outputs local JSON data files and interactive HTML reports but does not warn users that local workspace files may be created or modified. In agent environments, undisclosed persistence is dangerous because users may believe they are only getting conversational assistance while the skill silently writes artifacts, potentially overwriting existing files or exposing sensitive business data in generated reports.

Static analysis

No suspicious patterns detected.