ClawHub
Back to skill

Security audit

桌面自动驾驶

Security checks for vulnerabilities and agentic risk

Overview

This desktop automation skill is mostly aligned with its purpose, but it needs review because it can automate destructive GUI actions and persist raw typed user data in logs.

Review this before installing if you plan to use it on real desktop applications. Use it only with explicit per-task approval, avoid running bulk delete or submit workflows without manual confirmation, and treat its operation logs as sensitive because typed data may be stored in plain text.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Vague Triggers

Medium
Confidence
79% confidence
Finding
The skill uses broad activation wording such as invoking the skill whenever AI models, agent orchestration, or LLM applications are involved, without tight trigger constraints. In a desktop automation skill with exec/write/read tools, overbroad invocation can cause the agent to apply high-impact GUI automation in contexts the user did not clearly intend, increasing the risk of unintended clicks, data entry, or workflow execution.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
The examples include batch clicking of 'delete' icons, which is an irreversible or destructive action, without requiring confirmation, dry-run guidance, or warning text. In a GUI automation context, users may adapt example code directly, and mistaken image matching or window focus errors can trigger mass deletion or corruption of user data.

Ssd 3

Medium
Confidence
97% confidence
Finding
The operation log examples show raw typed input, including an email address, being written to persistent logs in plain text. In a desktop automation skill, typed content may include credentials, personal data, tokens, or business-sensitive information; storing that data under a predictable path materially increases the risk of local disclosure, lateral access by other processes, and accidental retention of secrets.

Static analysis

No suspicious patterns detected.