ClawHub
Back to skill

Security audit

Cloud Manager

Security checks for vulnerabilities and agentic risk

Overview

This skill is a cloud-storage manager with powerful file, permission, backup, migration, and deletion workflows, but its instructions are poorly scoped and include mismatched generic review behavior.

Review before installing. Use this only for explicit cloud-storage management tasks, and require manual confirmation before any permission change, migration, cleanup, deduplication, version-history pruning, scheduled backup, or command that touches OAuth credentials. Do not route generic analytics or code-review requests to this skill.

SkillSpector

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 (4)

Description-Behavior Mismatch

Medium
Confidence
96% confidence
Finding
The skill is presented as a cloud storage management tool, but its documented interface and JSON output describe a generic LLM review/scoring workflow instead. This semantic mismatch can cause an agent to invoke the skill for unrelated analysis tasks or mis-handle outputs, increasing the risk of unsafe tool use, data misrouting, and user deception.

Intent-Code Divergence

Medium
Confidence
94% confidence
Finding
The documentation claims cloud-manager-specific behavior, but the declared inputs and sample outputs are for generic scoring/audit results. Such contradictory I/O contracts can mislead an orchestrating agent into trusting incorrect semantics, producing incorrect automation decisions or exposing unrelated user content to an LLM-style review path.

Vague Triggers

Medium
Confidence
91% confidence
Finding
The invocation guidance says to use the skill for broad data analysis, reporting, statistics, and visualization, which overlaps with many unrelated requests outside cloud storage. In an agent environment with exec access, this over-broad scope can lead to inappropriate invocation on sensitive or irrelevant data, expanding the blast radius of the tool beyond its stated domain.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
The skill advertises cleanup, deduplication, migration, and version-history deletion behavior, including statements that some deletions are irreversible, but it lacks a prominent upfront safety warning and confirmation requirements. In a cloud-management context with exec capability and access to user storage, destructive operations can cause permanent data loss across multiple providers if triggered incorrectly.

Static analysis

No suspicious patterns detected.