ClawHub

MissionCore AI - Give It a Goal, It Executes Relentlessly

Security checks across malware telemetry and agentic risk

Overview

This skill is a goal-planning helper with broad focus instructions and local progress logging, but no evidence of hidden network access, credential use, destructive behavior, or privileged installation.

Install only if you want an assistant to enter a focused goal-execution mode. Use explicit, bounded missions; do not include secrets in mission text; and require confirmation before actions that spend money, delete or modify important data, publish content, affect accounts, or impact other people.

SkillSpector

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

Vague Triggers

Medium
Confidence
86% confidence
Finding
The activation phrases include highly generic patterns like "Mission: [your goal]" and "Execute mission: [goal]", which can easily appear in normal user prompts or surrounding content. This creates a prompt-collision risk where the skill may activate unintentionally, causing the agent to adopt a more rigid, goal-maximizing behavior without clear user intent or explicit mode switching.

Vague Triggers

Medium
Confidence
91% confidence
Finding
The template uses broad, imperative language such as 'MISSION: {{goal}}' and 'Execute immediately' without any scope, safety constraints, authorization checks, or domain limits. This can cause an agent to interpret arbitrary user input as a mandate for action, increasing the risk of unsafe, unauthorized, or policy-violating behavior if the goal is harmful or insufficiently specified.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
This function appends arbitrary entry content to a journal file without any notice, consent, or visible disclosure to the user. In an agent skill context, silent persistence of user-provided or user-derived content can create privacy and compliance risks, especially if sensitive prompts, secrets, or personal data are logged to disk unexpectedly.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal