Cloudsway ScaleBox Sandbox

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:26 PM UTC
• Version:
  1.0.5

Overview

This is a coherent helper for managing ScaleBox cloud sandboxes, with powerful but disclosed remote execution and file-transfer capabilities.

Install only if you trust ScaleBox and need cloud sandbox execution. Treat SCALEBOX_API_KEY as account authority, upload only files intended for a third-party cloud environment, disable sandbox internet access when unnecessary, review commands before execution, and download any needed results before terminating or deleting a sandbox.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• 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
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (2)

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The skill explicitly promotes running untrusted code and remote code execution, but it does not give a strong warning that code, files, secrets, and outputs are transmitted to and executed in a third-party cloud environment. In a security-sensitive agent context, omitting that warning can cause users to upload sensitive data or execute attacker-supplied code under a false sense of safety.

Missing User Warnings

Low

• Confidence: 82% confidence
• Finding: The documented terminate and delete commands are destructive and can irreversibly remove sandbox state, files, and resources, but the skill does not clearly warn about that consequence at the point of use. This increases the chance of accidental data loss or premature destruction of evidence/artifacts during an automated workflow.

VirusTotal

63/63 vendors flagged this skill as clean.