Sitecore's Developer MCP: Let the Orchestration Begin
Why the biggest opportunity for Sitecore's new Developer MCP isn't what it can do alone, but what it can do alongside everything else.

Sitecore just released its Developer MCP server in beta, and I've got to say, this is one of the more exciting developments I've seen for Sitecore developers in a while.
But probably not for the reasons you'd expect.
There's already plenty of excitement around agents being able to discover components, inspect content types, and interact directly with Sitecore. Those are useful capabilities, but let's not pretend capable AI agents couldn't already accomplish quite a bit of that by digging through serialized YAML files or interacting with existing APIs.
That's not what has me excited.
What interests me is what happens when we start connecting Sitecore to everything else involved in delivering software.
We're Not Just Writing Code
Think about everything that goes into building a feature on a typical Sitecore project.
The requirements are in Jira. The designs are in Figma. Your code is in GitHub. Your content models and component configurations are in Sitecore. You're bouncing between all of them, trying to make sure everything lines up before you've even written much code.
We've gotten pretty good at using AI to accelerate implementation, but there's still a lot of work happening around that implementation.
Now imagine an agent that can interact with those systems through their respective MCP servers.
You hand it a Jira story. It reviews the requirements, inspects the Figma design, examines existing patterns in the codebase, and determines what needs to happen in Sitecore. It helps establish the content model, register the component, build the frontend, and prepare the changes for review.
That's not something Sitecore's Developer MCP does on its own today, and it would require additional integrations and some thoughtful orchestration.
But we're starting to see the pieces come together.
And personally, that's a lot more exciting than watching an AI agent generate another React component.
What About the Projects We Actually Work On?
Let's be realistic. Most enterprise development isn't building shiny new applications from scratch.
We're working within existing implementations. Multiple sites, established content models, years of architectural decisions, and functionality that nobody wants to accidentally break.
This is where I see some particularly interesting opportunities.
Imagine you're tasked with extending an existing feature across multiple Sitecore sites. An agent could inspect the live component definitions and site availability, compare that information with the codebase, and help identify what needs to change.
Or imagine building a new feature where the agent can handle some of the repetitive setup in Sitecore, create development content, and work with separate testing tools to validate the implementation.
None of those individual activities are necessarily groundbreaking. We've had APIs, CLI tools, scripts, and other ways to automate this work for years.
But getting those capabilities into the same development workflow, without building and maintaining a custom integration for every project?
That's where things start getting interesting.
Just Because the Agent Can Do It Doesn't Mean We're Done
There's another part of this I'm particularly interested in exploring as the beta evolves.
Creating or updating something in Sitecore is one thing. Making sure that change fits into the rest of the development process is another.
What happens after the agent creates a content type or registers a component? Has the frontend been implemented? Have the appropriate tests been written? Are the changes accounted for in the project's source control and deployment process?
Those are important questions, and the answers are going to depend on how we build these workflows.
I'm not suggesting we hand an AI agent the keys to Sitecore and hope everything works out. That sounds like a great way to create an entirely new category of problems.
I'd rather see agents that can coordinate changes across systems, validate their work, and prepare everything for review, with appropriate controls around what they're allowed to modify.
The goal isn't to remove engineers from the process. It's to eliminate the busywork while keeping the engineering judgment that actually matters.
This Is Where I Want AI-Assisted Development to Go
I'm looking forward to getting hands-on with Sitecore's Developer MCP and seeing how these ideas hold up outside of hypothetical scenarios.
I don't expect a beta release to solve the entire development lifecycle, nor should anyone. But I do think this is an important piece of where AI-assisted development is heading.
We've spent a lot of time talking about how quickly AI can generate code. I'm much more interested in how it can reduce the friction between requirements, designs, implementation, configuration, and testing.
Because the biggest productivity gains might not come from generating the next component in 30 seconds. They might come from eliminating the twenty other things we had to do just to get that component built, configured, tested, and ready to ship.
Sitecore's Developer MCP is another instrument in an increasingly capable ensemble. Now let's see what happens when we get the whole orchestra playing from the same sheet of music.