April was heads-down month. No big announcements, no pivots — just shipping. Seven integrations covering content management, databases, and container infrastructure, plus a privacy feature that quietly solves a problem most people don't notice until it matters.
Here's what shipped.
TL;DR: April 2026 Update
- Your Data, Never Seen — tool responses can now be returned as a secure, downloadable file instead of passing through the AI's context window.
- Write and Publish, Anywhere — WordPress, Studio CMS, and Nuclino let AI draft, manage, and publish content without you opening a single editor.
- Query and Store, Smarter — Elasticsearch, Supabase, and MongoDB bring your search indices and databases into the same AI conversation.
- Your Cluster, Under Control — Kubernetes lets AI inspect pods, pull logs, manage deployments, and more — directly from your AI client.
✨ What's New 🚀
Your Data, Never Seen
MCP Express was built on a specific idea: AI should be able to do things with your data without necessarily seeing all of it. Until now, that applied to how connections were handled. This month, it applies to what comes back.
When a tool returns a response — a database query result, a list of records, a file export — that response would normally pass through the AI's context window on its way to you. The AI reads it, summarises it, and works with it. For most tasks, that's fine. For others — payroll data, client records, anything commercially sensitive — it isn't.
With secure file downloads, tool responses can now be returned directly to you as a downloadable file, bypassing the AI entirely. The AI knows the operation is completed. It doesn't see what was returned. You get the data. Nothing sits in a conversation log.
It's a small shift in how the pipeline works, but a significant one for anyone handling data they'd rather keep tight.
To enable this, open the tool you want to restrict your AI access to the data and tick the secure download option. That's the only change needed on your end.
Once enabled, here's how your AI hands the data back to you:
Learn more about why your AI shouldn't see every data and how MCP Express deals with it.
✨ Integrations 🚀
Write and Publish, Anywhere
Three new integrations this month for anyone managing content — whether that's a WordPress site, a modern CMS, or a team knowledge base.
WordPress and Studio CMS ship with the same set of publishing operations: list posts, fetch by slug, create, update, delete, and publish. Whether you're managing a client's WordPress site or running your own Studio CMS instance, your AI can handle the full content lifecycle without you opening a dashboard. Draft in conversation, publish on command.
Nuclino is where the knowledge base angle comes in. Your AI can now search and retrieve content from your Nuclino workspace directly — useful for pulling in process documentation, runbooks, or client context before responding, without you having to copy and paste anything across.
Query and Store, Smarter
Three integrations this month, covering search and your data layer.
Elasticsearch lets your AI store and search documents against your indices directly. If you're running Elasticsearch for logs, product search, or any kind of structured retrieval, you can now ask your AI to run those operations in plain language — no query syntax required.
Supabase brings your Postgres-backed projects into the same conversation. The integration covers the full range of what you'd expect from a Supabase workflow: execute SQL, list tables, manage migrations, handle branches, deploy edge functions, get logs, pause or restore projects, and generate TypeScript types. If you're actively developing on Supabase, your AI can now participate in that workflow end-to-end.
MongoDB rounds out the data integrations. Your AI can find documents, insert one or many, update or delete records matching a filter, and count documents across your collections. Standard document database operations, available in plain language from your AI client.
Your Cluster, Under Control
The Kubernetes integration this month covers the operations that matter most day-to-day for anyone managing containerised workloads. Your AI can list pods, namespaces, services, deployments, and ingresses — and describe any of them in detail. It can pull logs from a specific pod, create or delete pods, scale deployments to a specified replica count, and inspect ingress configurations.
In practice: a client pings you about a production issue while you're away from your desk. Open your AI client, ask it to pull logs from the relevant pod, describe the failing deployment, and draft the fix — without SSHing in or opening a separate dashboard.
Combined with secure file downloads, log data can come back to you as a file rather than through the AI's context — useful when logs contain sensitive runtime information you'd rather not expose.
Tell us what to build next
We're adding integrations every month based on what freelancers actually need. If there's a tool you rely on that isn't here yet,
drop us a line.
Try It Now
All seven integrations and secure file downloads are available today. The free tier covers everything you need to get started — no credit card required. Your first MCP server takes under 5 minutes to set up.
Create your free account
