Security audit

Cloud Architect Free

Security checks for vulnerabilities and agentic risk

Overview

This appears to be a cloud-architecture helper, but it asks for broad exec/write authority and gives inconsistent disclosures about cloud-account access and credentials.

Review before installing. Use it only for cloud-architecture planning, and do not let it run AWS CLI, Terraform, file writes, callbacks, or external API actions unless you explicitly intend that and are using read-only or tightly scoped credentials.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
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
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Intent-Code Divergence

High

Confidence: 97% confidence
Finding: The skill claims it does not directly access user cloud accounts, yet it includes AWS CLI commands that query live EC2 inventory, CloudWatch metrics, and Cost Explorer billing data. In an agent context with exec enabled, these examples can normalize or trigger real access to sensitive account metadata and spending information, creating a mismatch between user expectations and actual behavior.

Intent-Code Divergence

Medium

Confidence: 88% confidence
Finding: The file gives inconsistent guidance about credentials, first stating no extra API key is required and later saying some functions require platform API keys. This ambiguity can mislead users about what secrets may be needed or exposed, increasing the chance of unsafe credential handling or accidental execution under ambient credentials.

Description-Behavior Mismatch

Medium

Confidence: 82% confidence
Finding: The description mixes cloud architecture with unrelated UI, poster, and brand-visual design use cases, indicating scope confusion or templated content contamination. Overbroad or mislabeled capability descriptions are dangerous because they can cause the agent to activate this skill in inappropriate contexts and apply tools or instructions the user did not intend.

Vague Triggers

High

Confidence: 91% confidence
Finding: The activation text is vague and spans unrelated domains, which can cause the agent to match and invoke the skill far outside its intended purpose. Because the skill also advertises exec/write capabilities, overly broad triggering materially increases the risk of unnecessary command execution, file modification, or external calls in unrelated conversations.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The skill advertises file writing, command execution, and external API integration without clear user-facing warnings about side effects, required permissions, or data exposure. In a tool-enabled agent, this can lead users to underestimate operational impact and permit actions affecting local files, system state, or external services.

Static analysis

No suspicious patterns detected.