October 2025 Summaries

Apollo MCP Server 1.0 has been officially released, offering a robust solution for connecting GraphQL APIs to language model-based applications (LLMs) efficiently and securely. This server transforms GraphQL operations into Machine-Client Protocol (MCP) tools, allowing clients like Claude Desktop and Cursor to access and use these tools without additional coding. Key features include hot-reloading for rapid development, semantic search for efficient schema exploration, and persisted queries for production safety. The server also supports OpenTelemetry for enhanced observability and CORS for browser-based integration. With a strong emphasis on security and flexibility, Apollo MCP Server uses OAuth for authorization and supports YAML configuration for diverse deployment environments. By leveraging GraphQL's inherent advantages, such as type safety and declarative relationship handling, the server minimizes token usage and latency while simplifying agent logic. Apollo envisions expanding beyond GraphQL to become a universal orchestration platform for AI applications, with ongoing efforts to integrate other APIs and data sources seamlessly.

Graph artifacts, introduced in Public Preview by Apollo, offer DevOps teams versioned, immutable schemas for managing supergraph deployments with enhanced control and traceability. These artifacts address common challenges faced by teams, such as the inability to pin to a specific schema version, slow rollbacks, and missing audit trails by providing a unique SHA-256 digest for each version, ensuring stability and traceability. By integrating with Apollo Router v2.7 or later, teams can deploy, roll back, and audit schemas with confidence, using industry-standard OCI-compliant tools. This innovation facilitates seamless CI/CD integration, consistent fleet-wide deployments, and compliance-friendly audit trails, transforming how schemas are managed in production environments. The approach allows for promotion from staging to production without drift, instant rollbacks, and consistent schema versioning across distributed deployments, with plans for future improvements to enhance CI/CD support and provide more flexible tooling options.

At the GraphQL Summit 2025, Apollo GraphQL unveiled a suite of enhancements designed to accelerate graph development and support the evolving needs of AI-powered applications. Key highlights include the launch of Apollo MCP Server 1.0, which facilitates seamless integration of large language models (LLMs) with GraphQL APIs, and the introduction of GraphOS MCP Tools to assist developers using natural language interfaces. The platform has expanded its capabilities with features such as graph artifacts for version control, subgraph API keys for enhanced security, and connectors for efficient REST API integration. Additionally, the release of Graph Runtime APM Templates aims to improve monitoring and performance insights, while the Apollo GraphOS Operator for Kubernetes simplifies the management of GraphQL infrastructures. New iterations of Apollo Client and Apollo iOS are also introduced, featuring reduced bundle sizes and improved compatibility with modern programming practices. These advancements position Apollo as a leading force in API orchestration for AI agents, mobile, and web applications.

Apollo has introduced APM dashboard templates for Datadog, aimed at enhancing the observability of GraphOS Router performance with minimal setup time. These templates allow platform and SRE teams to quickly gain insight into the health of the GraphOS Router, supergraph, and subgraph by importing the dashboards into Datadog. The templates facilitate the identification of performance issues like latency spikes and error rates by providing a clear view of the system's operations, including first-class GraphQL error tracking and drill-down capabilities from supergraph to subgraph levels. They support best practices in observability by mapping spans and metrics with stable operation and resource names, and promoting GraphQL errors into APM error views. To manage observability costs efficiently, Apollo recommends maintaining low cardinality for resource names and employing tail-based sampling for high-throughput services, while ensuring consistent application of Datadog’s Unified Service Tagging. The initiative is designed to streamline the monitoring process and improve operational insights, allowing teams to correlate performance shifts with version rollouts and other activities efficiently.

Subgraph API keys have been introduced to improve security and streamline deployment pipelines by granting narrowly scoped permissions to specific subgraphs within an organization's graph, ideal for CI/CD workflows. Unlike the broader Graph API keys, which provide access to an entire federated supergraph and pose risks of unauthorized changes by different teams, subgraph keys restrict permissions to designated subgraphs, adhering to the security best practice of the "principle of least permissions." These keys are beneficial for autonomous teams managing their subgraphs, allowing them to perform necessary operations like running checks and publishing schema changes without affecting other areas. Available in GraphOS Standard and Enterprise plans, subgraph keys can be created using Rover or the Platform API, are immutable, and should be securely stored in a secret manager, facilitating safer and more controlled deployment processes.

Apollo iOS 2.0 introduces a major overhaul of the Swift client, designed to leverage Swift's modern concurrency model for enhanced performance, type guarantees, and developer experience in GraphQL-powered apps across Apple platforms. The update aligns with Swift 6's strict concurrency checks, offering APIs that ensure data safety across tasks, while simplifying workflows with async/await expressions instead of traditional completion handlers. This version requires migration to iOS 15, macOS 12, and newer platforms, and drops CocoaPods in favor of Swift Package Manager and XCFrameworks. While it currently supports only HTTP-based GraphQL subscriptions, an updated WebSocket implementation is forthcoming. The transition is structured to minimize risks, allowing developers to adapt their codebase progressively. A detailed migration guide is available to assist developers in adopting the new structured APIs and modernizing their client code.

Subgraph and Connector-level insights have been introduced in GraphOS Studio to provide developers with enhanced visibility into their specific APIs' performance within a federated GraphQL architecture. This new feature offers a dedicated dashboard that allows developers to pinpoint performance issues by identifying high-impact clients and operations, understanding error distribution, and analyzing latency and error patterns over time. By comparing metrics across different subgraphs, developers can determine whether performance issues are isolated or system-wide. The insights are powered by data collected from Apollo Router, which measures fetch latency and integrates with various APM tools. Available to all GraphOS Studio customers using Router 2.7.0 and above, the feature is part of a broader effort to democratize observability, with plans to make subgraph insights more accessible and extensible in future updates.

GraphQLConf 2025 in Amsterdam marked the tenth anniversary of GraphQL's open-sourcing by Meta, highlighting its evolution from a mobile application tool to a fundamental component of API orchestration for AI-powered applications. The conference showcased significant innovations, including Meta's new @async directive, advancements in query plan validation, and the introduction of input polymorphism via the @oneOf directive. GraphQL's role in AI development was a major theme, with discussions on making APIs AI-accessible, schema design patterns, and the evolving MCP protocol. The conference also underscored the maturation of GraphQL Federation as a standard infrastructure and the continuous enhancement of client tooling, emphasizing fragment colocation and caching capabilities. Community involvement was celebrated through new programs like GraphQL Locals and the GraphQL Ambassadors program. The conference concluded with excitement for the upcoming GraphQL Summit in San Francisco, where further exploration of GraphQL's integration with AI and its application in agentic development is anticipated.