What if you could just type, or say: "Publish a new blog post announcing our summer sale, use this photo, and schedule it for Friday at 9 AM" — to Claude, ChatGPT, or any AI assistant that supports MVC* — and have it actually happen?

That's exactly what our latest version does!

The problem it solves

For many of our clients, the backstage editor is powerful — but it's still a tool, and tools require context-switching. You have to remember where things are, how fields are named, and which settings matter. For a busy shop owner, a volunteer coordinator, or a small marketing team, that friction adds up over a week, a month, a year.

We remove that friction. It turns "I need to update the website" from a chore into a two-line message. No new interface to learn — you just describe your intent in plain language to an AI you probably already use for other things.

What you can actually do with it

Once connected, the AI has access to a set of purpose-built tools that mirror what our content editors do every day:

• Search and browse content — "What are our five most recent news articles?" or "Find the item about our Christmas event."
• Read a page in detail — pull up the full text, images, publish status, and URL of any item.
• Create new pages and articles — give it a title, some text (Bootstrap-friendly HTML is fully supported, so buttons, cards, and columns all just work), an author, a button link, and even a scheduled publish date.
• Upload images — hand the AI an image URL or paste one in, and it uploads it, ready to attach as a hero image, thumbnail, or background.
• Draft or publish immediately — every new item can be created as a draft for review or pushed live right away, your choice.
• Set SEO metadata — meta title and description can be set in the same request, so pages are search-ready from the moment they're created.

Because everything runs through your existing Exire Nexus services, anything the AI creates behaves exactly like content created through the backstage editor — same permissions, same structure, same categories.

A real workday, reimagined

Instead of:

Open laptop → log in → find "New Item" → fill in ten fields → find the image on your phone → upload it → set the publish date → save → double check it looks right...

It becomes:

"Create a new article titled 'Our Summer Opening Hours', use this photo as the header image, and publish it this Friday morning."

That's it. The AI handles the rest — searching for the right settings, uploading the image, and creating the page in the background — while you get back to running your business, coaching your team, or just enjoying your coffee.

Why this matters

We built Exire Nexus around the idea that the same core building blocks — content, products, events, people — should be flexible enough to serve every kind of need without forcing you to learn a dozen different tools. The MCP Server extends that same philosophy into how you interact* with the platform: through natural conversation, wherever you already are.

It's not about replacing the editor — it's about giving you another way in, one that fits into the way you already work and think, so managing your website feels less like an IT task and more like simply describing what you want, and watching it happen.

*What is MCP?

MCP stands for Model Context Protocol — an open standard that lets AI assistants like Claude connect securely to external tools and systems. Instead of the AI just talking about your website, it can actually act on it: creating pages, uploading images, searching content, and more — all through a conversation.

Exire Nexus ships with its own MCP Server built right in. Once you connect it to Claude (or any MCP-compatible AI client), your website becomes something you can manage by simply describing what you want.

Built with security front and center

Letting an AI make changes to your live website naturally raises the question: is this safe? We designed the MCP Server with that concern as the top priority:

• Per-user API keys — every request is tied to a specific authenticated user, so the AI only ever acts with that person's permissions, never more.
• OAuth 2.0 support — for clients like Claude.ai, the server supports a proper OAuth client-credentials flow, so there's no need to paste secrets into a chat window.
• IP allow-listing and rate limiting — the endpoint can be locked down to known IP addresses, and repeated failed authentication attempts automatically trigger a temporary block.
• Configurable, hard-to-guess endpoint path — the MCP address itself doesn't have to be predictable.
• Request size limits — protecting against oversized or abusive payloads.

In short: the AI can only do what you are allowed to do, and only after proving who it's acting on behalf of.