ClawHub
Back to skill

Security audit

Task Extractor

Security checks across malware telemetry and agentic risk

Overview

This skill is a task tracker, but it is unusually assertive: it can activate broadly, write task details to a persistent workspace file, and push the agent toward execution without clear approval boundaries.

Review before installing. Use it only if you want an assertive workflow that creates TASK_QUEUE.md in the workspace and may resume from it later. Avoid using it with secrets, personal data, or confidential project details, and require explicit confirmation before sends, deployments, configuration changes, or sub-agent delegation.

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 (2)

Vague Triggers

High
Confidence
91% confidence
Finding
The activation logic is intentionally over-broad: it triggers on ANY message with 3+ actionable items and even says 'If uncertain whether to activate: activate.' That can cause the skill to capture ordinary conversations or mixed requests unnecessarily, increasing unintended handling, persistence, and workflow hijacking risk rather than a narrowly scoped assistive behavior.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill persists extracted user request contents into workspace/TASK_QUEUE.md before any user-facing receipt and without disclosure or consent. Because user messages may contain sensitive instructions, credentials, personal data, or confidential project details, this creates a data-retention and privacy risk that expands exposure beyond the immediate interaction.

VirusTotal

62/62 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.