Changelog

Neon plugin for OpenAI Codex

The Neon Postgres plugin is officially available in the OpenAI Codex plugin directory. It adds the Neon MCP Server and Neon-focused Agent Skills to Codex, so you can create and manage Neon projects, branches, and databases from chat, run SQL and migrations, and get guided help on connections, branching, autoscaling, Neon Auth, and more.

Install the plugin

Codex CLI: If you do not have the CLI yet, install it and start codex:

npm install -g @openai/codex
codex

Then in Codex run /plugins, find Neon Postgres, and choose Install plugin. Complete any Neon sign-in or connection prompts.

For Codex app instructions, more detail on what's bundled, and example prompts, see Codex plugin for Neon. For an overview, see the Neon blog.

Snapshot API responses include storage size fields

The snapshot object in the Neon API now supports full_size and diff_size fields for monitoring snapshot storage.

• Manual snapshots expose full_size: the full logical size at the time of the snapshot.
• Scheduled snapshots: the first scheduled snapshot reports the full logical size via full_size. Subsequent snapshots report a diff_size value, which is the storage since the previous scheduled snapshot.

{
  "snapshots": [
    {
      "id": "snap-twilight-boat-an3a2yx2",
      "name": "production at 2026-04-17 09:57:07 UTC (manual)",
      "source_branch_id": "br-gentle-leaf-anax5tl3",
      "created_at": "2026-04-17T09:57:09Z",
      "manual": true,
      "full_size": 30965760
    }
  ]
}

The fields are supported on snapshot objects in responses from:

• Create snapshot
• List snapshots
• Update snapshot

For more on these fields (when each is present, omitted, or zero, and how that relates to charging and incremental billing), see Snapshot size fields in API responses in Backup & restore.

New NAT gateway IPs and VPC endpoint services in Asia Pacific (Singapore)

We've expanded infrastructure capacity in the AWS Asia Pacific (Singapore) region (ap-southeast-1) with new NAT gateway IP addresses and new VPC endpoint service addresses for Private Networking.

If you use Private Networking in ap-southeast-1, you can now use the additional VPC endpoint service addresses for enhanced capacity and reliability. See the Regions documentation for the complete list of NAT gateway IPs and the Private Networking guide for VPC endpoint service addresses by region.

Plugins tab for Neon Auth Organization settings

The Neon Auth page now includes a Plugins tab. Organization plugin settings (limits, creator role, invitation email) moved here from Configuration, so plugin-related options stay in one place.

See Organization plugin for what each setting does and how it maps to your app.

Upgrade your coding agent with the Neon Skill

skills.sh is a registry of reusable Agent Skills for compatible coding assistants. Add the Neon skill from neon-postgres on skills.sh so your assistant gets structured guidance for Neon Postgres: connections, branching, Neon Auth, APIs, CLI, and MCP.

Watch the demo below to learn more.

Other ways to install (including Cursor, Claude Code, npx skills, and neonctl init) are covered in Agent Skills.

Diagnose production errors with Sentry, Neon MCP, and database branching

Diagnosing and fixing production errors with Sentry and Neon MCP shows how to connect an AI agent (such as Cursor) to the Sentry and Neon MCP servers. You pull stack traces and the failing query from Sentry, then use Neon branching to create an isolated database copy, apply and validate fixes (for example with EXPLAIN ANALYZE), and ship the change to production when you are ready.

AI-assisted shortcuts in the Neon Docs

Copy page in the header opens a menu on every Neon Docs page: copy the page as Markdown, or open it in ChatGPT or Claude to ask questions with the page in context.

Set up Neon with AI appears in the right-hand sidebar. Choose it to open a modal where you can copy npx neonctl@latest init (npm or Homebrew tabs).

The init command gives your assistant Neon context and configures the Neon MCP Server. For the full flow and supported tools, see Connect MCP clients to Neon.

Postgres extension updates

neon extension version update

The neon extension is now version 1.14, up from 1.9. The neon extension can be used to monitor Local File Cache usage (how often pages are served from cache versus Neon storage) through the neon_stat_file_cache view and EXPLAIN ANALYZE. See The neon extension for details.

pg_search deprecation

Neon support for the pg_search extension is deprecated. As of March 19, 2026, it is not available for new Neon projects.

If you already use pg_search: you will continue to have access to the extension on your existing projects. Our team will contact you to discuss alternative options and deprecation timelines. You do not need to take action before we reach out. See The pg_search extension for alternatives you may want to explore.

Azure region deprecation

As of April 7, 2026, all Neon Azure regions (azure-eastus2, azure-westus3, and azure-gwc) are deprecated.

If your project already runs in an Azure region: your databases keep running as they do today.

If your organization actively uses Azure regions: you can continue to create new projects in Azure regions to maintain your current operations. Otherwise, new Azure project creation is no longer available.

We will contact you directly to discuss migration options and timelines. You do not need to migrate on your own before we reach out. See Region migration if you want to start planning a move.

Build stateful AI agents with Mastra and Neon

Building stateful AI Agents with Mastra and Neon Postgres shows how to use Mastra’s Memory module with Neon so agents keep context across threads and sessions instead of starting from scratch every time.

Build on Neon with the Vercel AI SDK

Build a Data-Driven AI Assistant on Slack with Vercel AI SDK and Neon Read Replicas walks through a Slack bot that answers data questions with the Vercel AI SDK. Queries run against a Neon read replica so production stays protected.