Fulcra Morning Briefing

• Outcome: Review
• Latest audit: May 29, 2026 at 12:00 PM UTC
• Version:
  1.1.7

Overview

This morning-briefing skill appears useful, but it handles highly personal context and sends location-derived weather requests to a third-party service without clear user-facing scoping or consent.

Install only if you are comfortable letting the skill process health, calendar, activity, and location-related information. Before using it, confirm which data categories it may read and whether weather lookups send your location to wttr.in; prefer coarse locations and disable external weather if you do not want that shared.

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
• Behavioral AST: exec() Call, eval() Call, Dynamic Import
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (4)

subprocess module call

Medium

• Category: Dangerous Code Execution

Content

def get_weather(location: str) -> dict[str, Any]:
    try:
        result = subprocess.run(
            ["curl", "-s", f"wttr.in/{location}?format=%l:+%c+%t+%h+%w"],
            capture_output=True,
            text=True,

• Confidence: 92% confidence
• Finding: result = subprocess.run( ["curl", "-s", f"wttr.in/{location}?format=%l:+%c+%t+%h+%w"], capture_output=True, text=True, timeout=10, c

Context-Inappropriate Capability

Medium

• Confidence: 95% confidence
• Finding: The skill reaches out to `wttr.in` even though it is presented as composing a briefing from Fulcra context. In a skill handling sleep, biometrics, calendar, and activity data, undisclosed third-party enrichment is risky because it violates expected data boundaries and may surprise users or reviewers about where data is sent.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The main behavior description normalizes collection of sleep, HR/HRV, calendar, and location-derived weather context before presenting a prominent upfront sensitivity warning or consent gate. Because these data types are highly intimate and can reveal health status, routines, meetings, and whereabouts, burying privacy constraints later in the document increases the chance an agent will over-collect or disclose sensitive information inappropriately.

Missing User Warnings

Low

• Confidence: 89% confidence
• Finding: User-supplied location data is transmitted to an external service without any user-facing disclosure or consent check in code. Location can be sensitive personal information, and in the context of a personalized briefing skill this creates a privacy issue even if no direct code execution occurs.

VirusTotal

66/66 vendors flagged this skill as clean.