AI protocols, such as MCP and Agent Skills, are agent-first, which risks bypassing the governance, security, and access controls that enterprises have spent years building around their APIs and data.

The New Stack is a publication covering trends and technologies in cloud-native development, DevOps, and software delivery. Developers can learn about containerization, Kubernetes, and cloud computing, as well as explore topics such as microservices architecture, serverless computing, and continuous integration/continuous delivery (CI/CD) pipelines.

The New Stack

Enterprise API landscapes have accumulated layers of competing standards (WSDL, Swagger/OpenAPI, AsyncAPI, Smithy, TypeSpec) driven by vendor interests and evolving architectural patterns. New AI-first protocols like MCP and Agent Skills risk bypassing the governance, security, and access controls organizations built around their APIs over decades. This creates data exposure risks as LLMs gain direct access to information repositories without proper boundaries. Organizations need comprehensive API mapping and governance strategies that bridge tactical AI needs with strategic enterprise requirements, rather than taking technology-first approaches that ignore business outcomes and security implications.

Your AI strategy is built on layers of API sediment