I had a client with about a dozen small WordPress sites—nothing major, mostly local businesses. Every month, they wanted a simple health check report. Traffic, comments, make sure nothing’s broken. Sounds easy, but it was a total mess. It meant logging into 12 different dashboards, clicking through the same screens, and copy-pasting stats into a spreadsheet. A mind-numbing, multi-hour task that I dreaded every single time.
My first thought was, naturally, to over-engineer a solution. I’ll just build a custom dashboard, pull everything in via the REST API. You know, the works. But the maintenance on that thing—handling auth for a dozen sites, building a UI, fixing it when a plugin update breaks an endpoint… it becomes its own project. A rabbit hole. The real problem isn’t getting the data; it’s the insane friction of securely accessing it across multiple sites. This is where a proper WordPress AI integration protocol is long overdue.
Beyond REST: Why MCP is the Right Tool
The core issue is context. The REST API is great for raw data, but it’s stateless. You have to teach your application what to ask for, every single time. It doesn’t inherently understand what “latest post performance” means. You have to make separate calls for the post, its analytics, comments, etc., and then stitch it all together. It’s a lot of boilerplate code for a simple question.
I saw an announcement from the WordPress.com team that they’re now supporting the Model Context Protocol (MCP), and it clicked. This isn’t just another API. It’s a standardized way for an application—or in this case, an AI assistant like Claude—to securely ask a WordPress site for information in a way that the site understands. It handles the OAuth handshake, so you’re not juggling API keys in a config file. Total nightmare averted right there.
Instead of building a custom dashboard, I can now just connect an AI tool that speaks MCP. It’s read-only for now, but that covers 90% of the reporting grunt work. It turns a multi-hour manual task into a single, well-formed question.
# This isn't real code, but conceptually, this is what MCP enables.
# Instead of multiple REST API calls, you ask a direct question.
$prompt = "Run a health check on all connected sites.
- Show me the latest post for each.
- Report traffic stats for that post (views, referrers).
- List any plugins needing updates.
- Summarize recent comments."
# The AI Assistant handles the individual MCP tool calls behind the scenes.
# No more manual logins. No more custom dashboard maintenance. Just the answer.So, What’s the Point?
This isn’t about letting AI write your posts. Not for me, anyway. It’s about automating the boring, repetitive parts of our job so we can focus on the hard problems. The stuff clients actually pay us to solve. Here’s the kicker: because MCP is an open standard, this is just the beginning. Self-hosted sites will get there. It’s a fundamental shift in how we interact with site data.
- Stop building one-off tools: Don’t build a custom dashboard for a problem a standard protocol can solve.
- Embrace secure standards: Using OAuth built into a protocol is infinitely better than managing a list of client API keys.
- Focus on efficiency: This frees up hours of non-billable (or low-billable) admin work. For real.
Look, this stuff gets complicated fast. If you’re tired of debugging someone else’s mess and just want your site to work, drop my team a line. We’ve probably seen it before.
Is this going to replace developers? No. But the devs who learn to use tools like this to automate their grunt work are going to leave the others in the dust. What repetitive task in your workflow would you kill with this?
Leave a Reply