Aws Cloud Architect Free

• Outcome: Review
• Latest audit: Aug 2, 2026 · 9:55 AM UTC
• Version: 1.0.2

Overview

This AWS skill is mostly a documentation-style assistant, but it claims to be read-only while showing commands that can create or modify real AWS resources and costs.

Review this skill before installing. If used, run it only with least-privilege read-only AWS credentials unless you intentionally want remediation or provisioning help, and do not allow an agent to auto-execute the example commands without manual review and confirmation.

SkillSpector

Vulnerability Patterns

• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Behavioral AST: exec() Call, eval() Call, Dynamic Import
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration

Findings (7)

Intent-Code Divergence

High

• Confidence: 98% confidence
• Finding: The skill repeatedly represents itself as read-only, yet it includes concrete mutating AWS CLI commands such as create-vpc-endpoint and put-retention-policy. In an agent context with exec/write tools enabled, this mismatch can cause operators or downstream automation to treat the skill as safe for inspection-only use while actually performing infrastructure changes that incur cost or alter security posture.

Intent-Code Divergence

High

• Confidence: 99% confidence
• Finding: The documented boundary says the skill only supports read-only operations, but the examples directly show resource creation commands like run-instances, create-db-instance, and create-vpc-endpoint. This is dangerous because users and orchestration systems may rely on the declared boundary when granting tool access, leading to unauthorized or unintended AWS resource creation and billing exposure.

Intent-Code Divergence

Medium

• Confidence: 95% confidence
• Finding: Recommending modify-db-instance --no-publicly-accessible contradicts the claim that the skill supports only read-only operations. Although the specific action improves security, it is still a state-changing command, so the inconsistency can bypass trust assumptions and cause an agent to execute modifications under a read-only label.

Vague Triggers

High

• Confidence: 90% confidence
• Finding: The description contains unrelated creative-design trigger text that does not match the AWS architecture purpose of the skill. Overly broad and mismatched invocation language can cause the wrong skill to be selected in automated routing, exposing exec-enabled AWS functionality in contexts where it was not intended and increasing the chance of accidental command execution or credential disclosure.

Vague Triggers

Medium

• Confidence: 88% confidence
• Finding: The invocation scope is ambiguous because the skill mixes AWS infrastructure assistance with unrelated broad usage contexts and generic automation claims. In agent ecosystems, ambiguous routing increases the chance that this exec-capable skill is invoked for inappropriate prompts, which can lead to unnecessary access to cloud credentials or unsafe command suggestions.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The markdown includes mutating AWS CLI examples near guidance text without a strong local warning immediately adjacent to those commands. Readers or agents may copy or execute the commands assuming they are safe examples, causing real infrastructure changes, retention-policy modifications, or cost-affecting network configuration changes.

Missing User Warnings

High

• Confidence: 99% confidence
• Finding: The architecture example shows resource-creating commands despite later asserting the skill is read-only, creating a sharp mismatch between stated safety boundary and executable guidance. In this skill's context, the danger is amplified because the declared tools include exec and write, making it plausible that an agent or user will run these provisioning commands and create billable or exposed AWS resources.

Static analysis

No suspicious patterns detected.