Security audit

Local event finder

Security checks across malware telemetry and agentic risk

Overview

This skill is a disclosed local event-finding assistant with optional calendar features that require user opt-in.

Before installing, review the optional setup choices carefully. Enable calendar access only if you want availability checks or one-click event creation, and enable the weekly automation only if you want recurring event scans. Periodically inspect or delete the skill's local runs and preference files if the machine is shared or synced.

SkillSpector

By NVIDIA

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
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

Findings (12)

Description-Behavior Mismatch

Medium

Confidence: 91% confidence
Finding: The skill metadata and description say it does not handle meeting planning, but the procedure explicitly includes adding suggested events to Google Calendar on request. This creates a scope mismatch that can cause the agent to perform side-effecting calendar actions when users or higher-level orchestration expect a discovery-only skill, increasing the risk of unintended writes to external systems.

Intent-Code Divergence

Medium

Confidence: 89% confidence
Finding: The documentation contradicts itself by excluding meeting planning while later instructing the agent to add events to Google Calendar. Such contradictions weaken policy enforcement and make it easier for orchestrators or users to invoke the skill outside its intended scope, especially where calendar modification is considered a privileged action.

Description-Behavior Mismatch

Medium

Confidence: 92% confidence
Finding: The setup step expands the skill from event discovery into calendar-tool installation and event addition, which exceeds the declared scope of a local event scanner. This broadening increases access to sensitive user data and write-capable actions without clear necessity, raising the risk of over-privileged behavior or unexpected side effects.

Description-Behavior Mismatch

Medium

Confidence: 91% confidence
Finding: The instructions add recurring automation creation, which is a persistent orchestration capability outside the stated purpose of simply discovering local events. That can cause the skill to create ongoing behavior the user may not expect from a one-time scanning tool.

Context-Inappropriate Capability

High

Confidence: 96% confidence
Finding: Attempting to enable a calendar connector grants access to an external account and potentially sensitive schedule data, which is not necessary for basic local event scanning. Because connector enablement expands privileges and may permit later write actions like adding events, it creates a meaningful overreach risk relative to the skill's declared function.

Context-Inappropriate Capability

Medium

Confidence: 89% confidence
Finding: Creating a weekly automation introduces an autonomous recurring action unrelated to the core event-scanning function and can persist beyond the immediate conversation. Persistent automations can generate unwanted prompts, notifications, or downstream actions if the user does not fully understand that an ongoing task is being created.

Missing User Warnings

Low

Confidence: 92% confidence
Finding: The README states that scan results are saved to the skill’s runs/ directory, but it does not prominently warn users that event searches, preferences, locations, and possibly derived recommendations may persist on disk. In a personal-assistant context, that can create unintended local data retention and privacy exposure, especially on shared machines or in environments where skill directories are synced, backed up, or accessible to other tools.

Vague Triggers

Medium

Confidence: 74% confidence
Finding: The invocation description is broad enough to match many generic requests about local activities, scanning, or shortlisting events without sharply defined exclusions. Over-broad triggers can cause unintended activation of the skill, leading to unnecessary data collection, user confusion, or execution of its workflow in contexts where another skill would be safer or more appropriate.

Vague Triggers

Medium

Confidence: 76% confidence
Finding: The phrase "what's happening in their area" is a vague everyday expression that may match casual conversation or broad recommendation requests. This increases the likelihood of accidental skill activation and can route user intent into a workflow that performs file writes and multi-step processing without sufficiently specific user intent.

Missing User Warnings

Low

Confidence: 80% confidence
Finding: The skill allows modifying user preference files without requiring a user-facing notice at the time of change. Even though the files are within the skill's configuration area, silent edits can undermine user expectations about what data is being stored or altered.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The instructions discuss enabling a calendar connector and adding events without a clear privacy and integrity warning about account access, calendar visibility, or write effects. Users may not appreciate that this could expose schedule data and permit changes to their calendar.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The setup flow omits a clear warning that creating a weekly automation establishes an ongoing background action rather than a one-time response. Without that disclosure, users may unknowingly authorize persistent behavior with future effects.

VirusTotal

59/59 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.