ClawHub

Lembrete Agua

Security checks across malware telemetry and agentic risk

Overview

This is a simple hydration reminder skill with disclosed personal tracking, but no code, credentials, file access, or hidden behavior.

Install this if you want a Portuguese hydration reminder that tracks simple personal details. Share only approximate weight, wake time, and water-intake entries you are comfortable keeping in the chat history, and be aware that the Goiânia heat guidance may not fit users in other locations.

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 (4)

Vague Triggers

Medium
Confidence
94% confidence
Finding
The activation phrases are broad enough to match ordinary conversation about hydration, which can cause the skill to activate unexpectedly and begin collecting health-related inputs such as weight and wake-up time without clear user intent. In a conversational assistant, overbroad triggers increase the risk of unintended state changes, confusing behavior, and privacy-sensitive data capture.

Vague Triggers

Medium
Confidence
95% confidence
Finding
The logging triggers include common words like 'bebi' and 'tomei', which are ambiguous in natural conversation and may cause the assistant to record water consumption when the user was referring to something else. This can corrupt stored health-tracking data and create misleading reports or recommendations based on inaccurate input.

Natural-Language Policy Violations

Medium
Confidence
84% confidence
Finding
The skill description hardcodes Goiânia-specific climate adaptation without indicating user choice, which can lead to incorrect location assumptions and unwanted personalization. While not directly a severe security flaw, it is a privacy and consent issue because environmental tailoring implies location-based inference or behavior without explicit opt-in.

Natural-Language Policy Violations

Medium
Confidence
90% confidence
Finding
Repeated instructions to tailor messaging to Goiânia and local heat conditions without opt-in reinforce an assumed location context throughout the interaction. This makes the skill more dangerous because it normalizes unsolicited personalization and may provide inaccurate hydration guidance if the user is elsewhere, while also creating privacy concerns around implicit location use.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal