Tool Claims to Catch Costly, Insecure AI-Generated AWS Architectures Before Deployment

A developer tool called the AWS Solutions Architect Prover has been released to address a recurring problem where AI coding assistants like Claude and Cursor generate overly complex, expensive, and insecure cloud architectures on AWS. The tool works as an MCP server that reviews proposed designs against five mandatory checks derived from the AWS Well-Architected Framework, rejecting any architecture that fails to meet defined criteria. Its creators argue that large language models default to pattern-matching from training data rather than applying cost or security reasoning specific to a user's actual requirements. Common AI design flaws cited include vague scalability claims without real performance targets, unaccounted costs like NAT Gateway fees and cross-AZ data transfer charges, and unnecessary service sprawl across tools like Kinesis, MSK, and EventBridge. Notably, the tool operates without requiring AWS credentials, positioning itself as a design-phase reviewer rather than an infrastructure automation agent.