ClawHub
Back to skill

Security audit

Cloud Storage

Security checks across malware telemetry and agentic risk

Overview

This cloud-storage skill is not clearly malicious, but its cloud-modifying authority is paired with inconsistent instructions and weak safety scoping.

Review before installing. Only use this with explicit, user-directed cloud-storage tasks, verify the exact provider/account/bucket before any operation, and avoid letting it handle SQL or database requests until the documentation is corrected and mutating actions require confirmation.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • 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 (4)

Description-Behavior Mismatch

High
Confidence
97% confidence
Finding
The skill metadata and documentation claim to manage cloud storage, but the same description says it should be used for database operations and SQL queries, and later shows an unrelated grading-style output. This semantic mismatch can cause the agent to invoke the skill in the wrong context, leading to unintended command execution or file/cloud actions on user data under false assumptions about what the skill does.

Intent-Code Divergence

Medium
Confidence
95% confidence
Finding
The documented output is a scorecard/report object unrelated to upload, download, listing, migration, or cost operations. This inconsistency can mislead orchestrators or users into trusting incorrect postconditions, causing unsafe automation decisions or masking real cloud-side effects.

Vague Triggers

High
Confidence
94% confidence
Finding
The trigger text is overly broad and inconsistent: a cloud-storage skill is declared applicable to database operations, SQL queries, and data storage management. In an agent environment, such broad routing criteria can cause unintended invocation of a skill with exec capability, increasing the chance of acting on the wrong resources or credentials.

Missing User Warnings

Medium
Confidence
90% confidence
Finding
The skill declares access to read and exec and is described as performing cloud/file operations, but it does not prominently warn that commands may run and that user data or cloud resources may be modified. In this context, missing safety disclosure increases the risk of users or agents triggering destructive or billable actions without informed consent.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.