轻量级主动代理

• Outcome: Review
• Latest audit: May 28, 2026, 4:45 AM
• Version:
  1.0.0

Overview

This skill is openly a background automation agent, but it gives broad unattended task, command, account, and data access without enough scoping or reviewable implementation detail.

Review carefully before installing. Only use this if you can verify the external `proactive-agent` implementation, run it with least privilege, keep automatic account-mutating actions disabled unless explicitly needed, and confirm visible start/stop controls, logs, notification settings, data-redaction behavior, and per-task approvals.

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 (4)

Vague Triggers

Medium

• Confidence: 93% confidence
• Finding: The skill description explicitly says it can run preset tasks automatically in the background without human triggering, but it does not define scope, authorization boundaries, or required user opt-in. For an autonomous agent skill, vague always-on activation language increases the risk of unintended actions, privilege misuse, and silent system-state changes.

Vague Triggers

Medium

• Confidence: 95% confidence
• Finding: These lines describe unattended periodic execution, monitoring, data sync, backups, and custom script/command execution without any clear invocation constraints or safety gates. Because the skill supports arbitrary command execution on a schedule, weakly specified controls make misuse or accidental destructive actions materially more likely.

Missing User Warnings

High

• Confidence: 97% confidence
• Finding: The skill advertises autonomous background execution of commands and operational tasks but does not provide a prominent warning that these actions can modify files, services, accounts, or other system state without immediate user awareness. In this context, omission of a strong warning is dangerous because the feature set includes custom commands, synchronization, backups, and unattended operation.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The skill describes checking mail, calendars, to-dos, and sending multi-channel alerts, but it does not warn users that this may involve access to sensitive personal or business data and transmission of that data to external notification channels. In a monitoring/alerting skill, this context increases privacy and data-leakage risk, especially if alerts include message content, schedules, or metadata.

VirusTotal

VirusTotal findings are pending for this skill version.