August 2025 Summaries

In July, the MCP Server Builder community held two significant meetups in San Francisco and New York, engaging engineers, toolmakers, and AI developers in discussions and demonstrations on building with the Model Context Protocol (MCP) for large language models (LLMs). Key sessions included Nina Lopatina's exploration of dynamic server selection using Contextual AI, Annie Wang's demonstration of Google's Agent Development Kit for simplifying the agent lifecycle, and YK Sugi's presentation on AI-driven UI testing with MCP servers. Michael Watson of Apollo showcased the efficient management of MCP servers with tools like hot-reloading, while Diptanu Gon Choudhury highlighted the protocol's role in scalable, stateful workflows. Other sessions covered using Postman's Flows for DevOps automation, securing MCP servers with zero trust principles by Nick Taylor, and voice-to-voice agent interactions by Hermes Frangoudis. The events highlighted the potential of MCP in enhancing agentic workflows and emphasized community collaboration to further develop and implement advanced MCP tools.

Apollo MCP Server has introduced authorization support, making it production-ready for AI integration by implementing a comprehensive OAuth 2.1 system with full PKCE support. This update addresses the need for secure access and identity verification, enabling teams to deploy AI agents in production while maintaining control over data access through a declarative approach using GraphQL. The new authorization feature allows for consistent security policies and identity enforcement, preventing unauthorized access and attack vectors. Alongside this, a shift from command-line arguments to YAML configuration files has been made to improve manageability and security. This development follows the ratification of the MCP Authorization specification and is aimed at enhancing the deployment capabilities of AI agents with robust security measures.

Apollo has announced the public preview of the Apollo GraphOS Operator, a tool designed to manage subgraphs and supergraphs as Kubernetes-native resources. This operator introduces three new Kubernetes Custom Resource Definitions (CRDs) that enable users to define subgraphs, compose them into supergraphs, and deploy these configurations using native Kubernetes resources and workflows. By leveraging the Kubernetes ecosystem, the Apollo GraphOS Operator provides end-to-end declarative control over subgraphs and supergraphs, ensuring schemas and implementations remain synchronized and enabling controlled deployments. The operator automatically discovers subgraphs, composes them into a supergraph schema, and continuously monitors the composition status, integrating seamlessly with existing Kubernetes capabilities such as rolling updates and monitoring tools. It simplifies the management of GraphQL supergraphs by utilizing GitOps-friendly methodologies and promises future enhancements like ephemeral supergraphs and safe deployment strategies.

On August 5, 2025, Apollo announced the public preview of its new GraphOS Operator, designed to simplify the management of GraphQL deployments within Kubernetes environments. This tool aims to streamline the development process by providing out-of-the-box best practices for managing GraphQL, reducing the effort required for setting up progressive deployments, rollbacks, and ephemeral development environments. It offers teams the flexibility to manage their GraphQL operations either declaratively, through Kubernetes resources, or via UI-based management with GraphOS Studio, aligning with their preferences. The operator supports cloud and on-premise Kubernetes deployments and integrates with popular observability and deployment tools like Datadog and Argo CD. The introduction of this operator is a step towards Apollo's vision of making GraphQL operations in Kubernetes more efficient and robust, with future updates expected to enhance capabilities such as canary deployments, fast rollbacks, and self-healing router instances.