March 2026 Summaries

AI agents are increasingly being used as interface layers for applications, allowing users to interact with their data through natural language rather than traditional dashboards and forms. Nhost has introduced an open-source service, the Nhost MCP Service, which enables any Nhost project to become an agentic service by exposing GraphQL APIs through the Model Context Protocol (MCP). This protocol allows AI assistants to authenticate via Nhost Auth and interact with data while respecting defined permissions, without requiring custom integration code. The MCP server, which sits between the AI assistant and the GraphQL API, provides tools for schema discovery and executing queries and mutations. The service relies on OAuth2 for authentication, and permissions are enforced based on JWT claims. A reference implementation of the service is available at mcp.nhost.io, and the setup process involves enabling OAuth2, building a consent page, and deploying the MCP service. Nhost encourages feedback from users to evolve the service, with plans to expand permissions and capabilities.

Nhost has introduced several key updates and improvements to its platform, including the addition of Nhost Auth as an OAuth2/OIDC provider, allowing projects to issue tokens to third-party applications similar to Google or GitHub. The platform now supports full event and cron trigger management directly from the dashboard, eliminating the need for the Hasura console, and has enhanced its database capabilities to include views, materialized views, and enums accessible via the GraphQL API. Data browser improvements include column-level filtering and customizable column visibility and ordering. The dashboard now features tools for managing GraphQL metadata, and support for PostgreSQL 18.1 has been added while dropping support for versions 14 and 15. Significant memory optimizations have been implemented in the storage service to improve stability, and the CLI is now installable via Homebrew and Nix. Nhost documentation is now embedded in the CLI, allowing for real-time access to up-to-date documentation, which enhances the performance of LLMs. The documentation site has been updated and rebuilt on Astro, offering a faster and more customizable experience. These updates, along with other behind-the-scenes improvements, contribute to a more reliable and user-friendly platform.