ClawHub
Back to skill

Security audit

图像

Security checks for vulnerabilities and agentic risk

Overview

This image-generation skill is mostly coherent, but it gives broad automation and sharing instructions without enough scoping or user confirmation.

Install only if you are comfortable with a skill that can automate the browser and desktop, read and copy local image files, and send a selected file to Feishu. Before use, confirm the exact generated file path and recipient/channel, and do not rely on the artifact's sandbox or whitelist language as a technical guarantee.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Intent-Code Divergence

Medium
Confidence
93% confidence
Finding
The skill claims command execution is restricted to a safe sandbox or whitelist, but the documented workflow uses unrestricted shell commands such as uvx desktop-agent, ls, cp, and sleep with no enforcement mechanism shown. This mismatch can mislead operators into over-trusting the skill and increase the chance that future edits or copied patterns introduce unsafe command execution under a false assumption of safety.

Vague Triggers

Medium
Confidence
83% confidence
Finding
The invocation language is extremely broad, including generic productivity and workflow-improvement scenarios, which can cause the skill to activate in contexts unrelated to image generation. Because the skill has read, write, and exec capabilities, accidental invocation expands the chance of unnecessary browser automation, local file access, and outbound sharing actions.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill instructs sending generated files to Feishu using a messaging tool but does not require an explicit user-facing consent step or warn that data will leave the local environment. In a tool with file access, this creates a real risk of unintended external data transmission, especially if the wrong file path is selected or if generated content includes sensitive information.

Static analysis

No suspicious patterns detected.