Security audit

FamilyWall

Security checks across malware telemetry and agentic risk

Overview

This FamilyWall skill appears purpose-aligned, but it handles sensitive family data and can write downloaded media to loosely controlled local paths.

Review before installing. Use this only if you are comfortable giving an agent access to private FamilyWall household data, including locations, messages, lists, recipes, and media. Avoid broad prompts that could trigger it accidentally, and only download media into a dedicated folder after checking filenames and overwrite risk.

SkillSpector

By NVIDIA

Vulnerability Patterns

Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
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
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (5)

Vague Triggers

Medium

Confidence: 86% confidence
Finding: Several triggers such as 'where is', 'recipe', 'download photo', or 'add to list' are broad enough to match ordinary conversation and unintentionally activate a skill with access to sensitive family data, messaging, and downloads. In this context, accidental invocation is more dangerous because the skill can reveal locations, private messages, or write local files.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The documentation advertises access to highly sensitive family content, including locations, messages, and media downloads, without explicit privacy warnings or notice about local file writes. Users may not understand that invoking the skill can expose household PII or store attachments on disk, increasing the risk of inadvertent disclosure.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The media download function writes server-fetched content directly to an arbitrary caller-supplied output_path with no path validation, sandboxing, or overwrite protections. In an agent context, a prompt or downstream tool invocation could cause files to be written outside an intended workspace, enabling clobbering of local files or planting untrusted content on disk.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The media download command writes attacker-controlled or remote-content-derived files to local disk without an explicit safety prompt or guardrails. Because the output path can be influenced by a direct URL-derived filename or user-supplied output value, this creates an unexpected filesystem side effect and increases the risk of overwriting files or placing untrusted content on disk.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The bulk download path creates directories and writes multiple files to disk based on remote metadata such as media names, with no explicit user warning or strong path sanitization. In the context of a family messaging/media skill, downloading attachments is expected functionality, but automatically materializing many remote files locally still carries meaningful risk from untrusted filenames, overwrites, and storage abuse.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.