MCP servers promise AI agents will conquer AWS and Kubernetes, but end up as abandoned GitHub repos forcing devs back to trusty CLIs

In the rapidly evolving field of AI tooling, the Model Context Protocol (MCP) has emerged as a promising standard for enabling Large Language Models (LLMs) to interact with infrastructure. However, despite its potential, MCP servers are currently being outperformed by Command Line Interfaces (CLIs) in managing platforms with complex REST APIs, such as AWS or Kubernetes. The implementation of MCP introduces friction, including difficulties in discovering and configuring servers, lack of shared configuration, and incomplete wrappers around existing APIs. In contrast, CLIs offer a mature and reliable solution, with a single binary providing nearly 100% coverage of a platform's REST APIs. With CLIs, authentication, transport, and upgrading are handled natively, reducing cognitive load and tool hunting errors. As the industry continues to develop AI protocol standardization, CLIs remain the superior choice for agentic workflows, offering faster execution, lower maintenance, and higher reliability.