ClawHub
Back to skill

Security audit

自动驾驶流程

Security checks across malware telemetry and agentic risk

Overview

The skill is a legitimate automation guide, but it encourages agents to build or enable automations too proactively, including scripts, cron jobs, file moves, and email actions, without consistently requiring explicit user approval.

Install only if you want an agent to help design and implement real automations. Before letting it run commands, require the agent to describe affected files, accounts, schedules, and external recipients; use dry-run first; and approve any cron job, watcher, email sending, or production-data write explicitly.

SkillSpector

By NVIDIA
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

High
Confidence
95% confidence
Finding
The skill explicitly instructs the agent to proactively build automation whenever it notices repetition, without requiring clear user confirmation, scope checks, or safety gating. In a skill with read+exec capabilities, this can lead to unauthorized script creation, scheduling, file modification, or outbound actions based on weak signals rather than explicit consent.

Vague Triggers

High
Confidence
96% confidence
Finding
The trigger signals treat common conversational patterns—such as a third similar request or similar pasted data—as sufficient basis to propose or initiate automation. These cues are too broad and can cause the agent to overreach from routine assistance into operational changes, especially when paired with execution tooling and examples that create persistent automations.

Missing User Warnings

Medium
Confidence
91% confidence
Finding
The implementation section provides executable patterns for scripts, cron jobs, file writes, and email sending, but it does not consistently foreground that these actions can alter local data, create persistent scheduled tasks, or affect external systems. Users may not appreciate that following the skill can produce real-world side effects beyond analysis, increasing the risk of unintended changes or data leakage.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.