yoooclaw-meeting-preparations

• Outcome: Review
• Latest audit: Jun 16, 2026 at 8:12 PM UTC
• Version:
  1.0.0

Overview

This meeting-prep skill is coherent, but it can read sensitive phone notifications and auto-install a web-search dependency without clear user confirmation.

Install only if you are comfortable with an agent searching recent phone notifications for meeting context. Before using it, confirm which people, topics, apps, and dates are in scope, and do not let it auto-install the byted-web-search dependency unless you trust that source and approve the environment change.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Context-Inappropriate Capability

Medium

• Confidence: 97% confidence
• Finding: The skill instructs the agent to run shell commands to inspect installed skills and to install a new package automatically via npx. That exceeds the core briefing task and creates a supply-chain and arbitrary-command execution risk, especially because the package source is remote and the action is performed without explicit user approval.

Vague Triggers

Medium

• Confidence: 87% confidence
• Finding: The activation criteria are broad enough to trigger on many ordinary meeting-related requests, which can cause the skill to run in situations where the user did not clearly ask for notification mining or web enrichment. In this skill, overbroad triggering materially increases the chance of unnecessary access to sensitive notifications and external lookups.

Missing User Warnings

High

• Confidence: 98% confidence
• Finding: The skill is designed to inspect recent phone notifications and combine them with web search results, but it provides no user-facing notice or consent step for accessing privacy-sensitive local data. This can expose personal or confidential communications unexpectedly, especially when triggered by a generic meeting-prep prompt.

Ssd 3

Medium

• Confidence: 96% confidence
• Finding: The instructions direct the agent to gather messages about named individuals from recent notifications and use them in the briefing, without any privacy screening, sensitivity classification, or minimization. That can surface private conversations, confidential work updates, or unrelated personal data merely because a person or keyword matched.

Ssd 3

Medium

• Confidence: 98% confidence
• Finding: The output template explicitly requires preserving the original meaning of messages and attaching source app, person name, and date, which increases the chance of directly leaking personal or confidential notification content into the final response. Combining attribution metadata with content makes re-identification and sensitive inference easier and more damaging.

VirusTotal

64/64 vendors flagged this skill as clean.