ClawHub

Sub Churn

Security checks across malware telemetry and agentic risk

Overview

This is a subscription churn planning skill that provides templates and guidance, with no code execution, credentials, persistence, or direct account access.

Before installing, treat generated email and SMS content as draft customer communications. Review discounts, pause or downgrade offers, cancellation wording, and regional subscription rules before sending, and avoid sharing unnecessary sensitive customer data when providing churn exports or examples.

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 (2)

Vague Triggers

Medium
Confidence
92% confidence
Finding
The manifest description includes broad natural-language triggers such as "subscriptions are leaking" and "people say it's too pricey," which can match ordinary business discussion rather than a clear request for this specific skill. That increases the chance of unintended invocation, causing the agent to apply subscription-retention workflows in contexts where the user may have wanted general analysis or a different skill, leading to confused actions or inappropriate guidance.

Vague Triggers

Medium
Confidence
95% confidence
Finding
The evaluation set includes an unrelated prompt about summarizing photosynthesis, which is broad and non-domain relative to a subscription-churn skill. This can train or validate overly permissive routing behavior, causing the skill to trigger on ordinary educational or generic requests instead of subscription-retention scenarios, increasing the risk of misrouting and inappropriate outputs.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal