Security audit

Water Tracker

Security checks across malware telemetry and agentic risk

Overview

The skill is a coherent local water tracker, but it can silently save health-related hydration details from broad casual mentions without a clear consent or deletion flow.

Review this before installing if you do not want casual drink, symptom, illness, medication, condition, weight, exercise, or alcohol references saved locally. The skill does not show evidence of uploading data or using credentials, but users should know it is designed for silent local tracking and should have a clear way to review or delete ~/Clawic/data/water/.

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 (6)

Vague Triggers

Medium

Confidence: 90% confidence
Finding: The skill advertises activation on very broad, common conversational cues such as casual drink mentions and nonspecific symptoms like headache, fatigue, thirst, or dark urine. That can cause unintended invocation in unrelated contexts and lead the agent to begin logging health-related data or giving hydration guidance when the user did not explicitly request it, creating privacy and safety risks around medical-adjacent advice.

Vague Triggers

Medium

Confidence: 94% confidence
Finding: The default trigger for 'any other hydration mention' is especially broad because it instructs the skill to silently log with timestamp and estimated volume even when the user only makes a passing reference. In practice, this can capture incidental conversation as persistent personal health data and may override user expectations about consent, while also increasing the chance of inappropriate advice in ambiguous situations.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The file explicitly instructs the agent to store declared medical conditions in memory, which involves sensitive health data. Without any minimization, consent, retention, or privacy-handling guidance, this creates a real risk of unnecessary collection and persistence of regulated or highly sensitive personal information.

Vague Triggers

Medium

Confidence: 81% confidence
Finding: The instruction says advice should appear only when asked or when a 'Quick Reference row triggers it,' but that trigger is undefined in this file. Ambiguous behavioral triggers can cause the agent to provide advice inconsistently or at unintended times, which is risky in a health-tracking context because users may receive unsolicited or omitted guidance.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The skill directs the agent to read and create persistent user data files containing hydration history, health context, weight, and patterns without any explicit user-facing notice or consent step. This creates a privacy risk because sensitive personal and health-related information may be accessed and stored silently, exceeding user expectations and weakening informed consent.

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: The file explicitly promotes 'silent and instant' logging and says the first drink mention should be logged without comment. Ongoing silent recording of health-adjacent behavioral data is dangerous because users may not realize persistent tracking is occurring, undermining transparency and consent.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.