ClawHub
Back to skill

Security audit

现场检查报告辅助技能

Security checks for vulnerabilities and agentic risk

Overview

This skill coherently creates local HTML site-inspection reports from user-provided notes and photos, with no evidence of hidden network access, credential use, or destructive behavior.

Install only if you are comfortable with inspection photos and notes being saved locally and embedded into a single HTML report. Review or delete inspection_images, image_data.json, and generated report files after use, especially when images show private facilities, personnel, or safety issues.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
95% confidence
Finding
The skill instructs the agent to save user images locally, generate JSON output, and write an HTML report file, but it does not declare permissions or clearly constrain those file-write operations. Undeclared write capability is dangerous because it can cause silent persistence of user-provided data, unexpected filesystem modification, and weak oversight of where sensitive inspection material is stored.

Vague Triggers

Medium
Confidence
90% confidence
Finding
Using the broad trigger word “完成” can cause accidental state transitions into report-outline generation during ordinary conversation, especially in multilingual or conversational contexts where that term is common. Premature report generation can lead to incomplete, incorrect, or unintended processing of user data and may trigger downstream file operations before the user actually intends to finalize the session.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The skill stores user images, invokes a local processing script, and writes an HTML output file, but it does not warn about data persistence, local file creation, or the sensitivity of inspection content. In a real enterprise inspection context, these files may contain confidential operational details, personally identifiable information, or evidence images, so silent storage materially increases privacy and data-handling risk.

Missing User Warnings

Medium
Confidence
97% confidence
Finding
Embedding photos as base64 directly into HTML makes the final report self-contained, which also makes sensitive images easy to copy, forward, archive, or exfiltrate without separate attachment controls. In the context of site inspection reporting, the images may reveal internal facilities, safety defects, personnel, or other sensitive details, so the self-contained format increases the blast radius of accidental sharing.

Static analysis

No suspicious patterns detected.