Geo Cycle Autopilot

• Outcome: Review
• Latest audit: Jun 3, 2026, 11:37 AM
• Version:
  1.0.0

Overview

This skill matches its GEO automation purpose, but it stores an API key in plain text and can perform scheduled SaaS/network actions without per-run approval.

Install only if you are comfortable with a scheduled agent reading a locally stored GEO API key, contacting ai.gaobobo.cn and source URLs, writing export ZIPs under your home directory, and updating GEO task status automatically. Prefer using a restricted/revocable API key and protect or replace the plaintext key file with a safer credential mechanism if available.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• 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
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (4)

Vague Triggers

Medium

• Confidence: 92% confidence
• Finding: The skill is explicitly designed for unattended daily loading and execution, but it lacks clear trigger boundaries, consent prompts, or safety gates before performing privileged actions. In context, that means network calls, local file reads, and exports can happen automatically, increasing the chance of unintended data access or repeated actions without user awareness.

Missing User Warnings

Medium

• Confidence: 97% confidence
• Finding: The skill instructs the agent to read, solicit, and persist an API key in a local plaintext file, but provides no meaningful privacy warning, secure storage guidance, or handling restrictions. This is dangerous because it normalizes credential collection and long-term local storage in a way that could expose secrets to other local processes, backups, logs, or future agent actions.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The skill defines automatic scheduled behavior that performs remote requests, content fetching, content generation workflow steps, and local ZIP exports without a clear user-facing warning about these side effects. In an autopilot context, silent network and filesystem activity is riskier because it can surprise users, consume resources, or handle sensitive content without contemporaneous approval.

Ssd 3

High

• Confidence: 99% confidence
• Finding: The skill explicitly tells the agent to ask the user for an API key, save it to disk, and reuse it later. This is a strong secret-handling anti-pattern because the agent becomes part of the credential collection and storage path, raising the risk of accidental disclosure, insecure persistence, and misuse during later autonomous runs.

VirusTotal

VirusTotal findings are pending for this skill version.