Security audit

Family Partner is an AI-driven family intelligent assistant, committed to becoming an indispensable digital partner for every family.We not only help families manage trivial affairs, but also understand family dynamics, promote member connections, and witness family growth.

Security checks across malware telemetry and agentic risk

Overview

This family-memory skill appears purpose-aligned, but it stores and may automatically resurface sensitive household and medical details with broad triggers that need review before installation.

Review before installing. Use it only if you are comfortable keeping sensitive family details, including allergies or medical notes, in a local SQLite memory store. Avoid entering highly sensitive child, health, school, or relationship details unless needed, and ask the publisher to add explicit confirmation before saving memories and before showing personal briefings from broad greetings.

SkillSpector

By NVIDIA

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Intent-Code Divergence

Medium

Confidence: 79% confidence
Finding: The README claims minimal access limited to ~/.openclaw/family-partner/, but backup examples write to ~/backups/, which expands the operational filesystem scope beyond the declared boundary. This mismatch can normalize broader file access for the skill/operator and weakens the least-privilege model expected by users and reviewers.

Missing User Warnings

Medium

Confidence: 82% confidence
Finding: The README promotes automatic reminders, milestone tracking, and family-memory features involving household routines, children, allergies, and anniversaries, but the early feature description does not prominently warn that sensitive personal data will be stored and resurfaced automatically. This can lead users to input highly sensitive family data without fully understanding retention and exposure risks on shared devices or interfaces.

Vague Triggers

Medium

Confidence: 89% confidence
Finding: Using a generic phrase like "good morning" as a skill trigger can cause the morning-briefing workflow to activate when the user is simply greeting the assistant, leading to unintended disclosure or access of stored family data. In a skill that aggregates schedules, tasks, anniversaries, and other household information, accidental activation increases privacy risk even without malicious intent.

Vague Triggers

Medium

Confidence: 93% confidence
Finding: The trigger phrase "remember..." is overly broad and may capture ordinary conversational text, causing the assistant to persist information without sufficiently clear user intent. Because this skill stores sensitive family and medical details, accidental writes can create privacy, retention, and data-integrity problems.

Missing User Warnings

High

Confidence: 97% confidence
Finding: The skill explicitly supports storage of allergies, medical, school, work, and family preference data, but does not clearly warn users that this sensitive information is persisted in a local SQLite database. Lack of transparent disclosure undermines informed consent and increases the chance that highly personal household data is stored without the user's understanding of retention and exposure risks.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.