When AI Writes the Web: A New Age in WordPress Development
It started with a question: Can an AI not just write content, but build the entire infrastructure that hosts it?
The answer—demonstrated in a live, fully-functional website—is yes.
Letters from the Ice (https://dont.histick.us/letters) isn’t just another WordPress blog. It’s a real-world case study showing how the Model Context Protocol (MCP), combined with the AI Engine WordPress plugin, enables generative AI to build, design, and maintain a site from scratch—with very little human interference.
Let’s unpack how this was done, what tools were used, and how website owners can replicate this groundbreaking process.
The Engine Behind the Ice: What is MCP?
Model Context Protocol (MCP) is a communication framework that allows AI models to interact directly with external tools and websites. Think of it like giving an AI permission to act—not just talk.
With MCP, a WordPress site becomes more than a CMS. It becomes a programmable environment where AI can:
- Write and edit code
- Install and configure plugins
- Modify themes
- Generate and manage content
- Collaborate with other models or agents
Key Features of MCP:
- Tool-Based Permissions: Only Tools explicitly enabled on the MCP server can be used by the AI, ensuring control and safety.
- Task-Oriented Execution: AI receives context and goals, then executes tasks like a developer would.
- Cross-Agent Collaboration: Multiple AI models or agents can connect, share context, and collaborate on tasks.
The AI Engine Plugin: The Gateway to Orchestration
1. MCP Server Mode
Turn WordPress into an MCP server. Expose only the Tools you choose (create_post, upload_media, etc.) so external AIs—like Claude Desktop—can safely update content inside a controlled sandbox.
2. Orchestration
Let your WordPress‑based bot reach out to other MCP servers. It can chain Tools across services for complex, real‑time workflows that go beyond WordPress.
A lightweight connector script, mcp.js, handles the handshake. Once linked, the AI stops merely advising and starts doing—creating posts, uploading media, and more, all on command.
Inside the Ice: How Letters from the Ice Was Built by AI
The site Letters from the Ice is a living example of AI-enabled development using MCP. Here’s how the process unfolded.
Phase 1: Setting Up the Infrastructure
- A Pro Suite WordPress hosting account was provisioned.
- The AI Engine plugin was installed and configured to run in MCP Server mode.
- A Claude Desktop Application was connected via `mcp.js`, enabling direct orchestration.
Phase 2: Prompt Engineering & Planning
Human input came in the form of structured prompts, such as:
Phase 3: Theme Creation and Design
Using the MCP Toolset:
- AI created a custom WordPress theme from scratch.
- Included minimalist CSS and custom templates for letter-style posts.
- Handled theme installation and activation autonomously.
Phase 4: Plugin Development
A fully functional custom plugin was generated by AI to:
- Register a new post type (`Letter`)
- Add custom fields like “Time Period,” “To:,” and “From”
- Provide a backend UI for admins to manage these fields
Phase 5: Visual Content with WP AI Agent
The AI also interacted with MyCelf’s WP AI Agent plugin, a premium content and image generation tool accessible via custom registered MCP Tools.
- AI generated images
- Media was automatically uploaded and assigned where necessary
Phase 6: Ongoing Updates and Maintenance
Claude, still connected via MCP, listens for commands or scheduled tasks:
- Update themes/plugins
- Add new content
- Refactor layouts based on new prompts
- Implement a ‘Scroll to Top’ mechanism to enhance UX/UI
Why This Matters: The Future of AI-Driven Development
This case study proves that AIs can now build, not just write. With tools like MCP, Claude, and the AI Engine plugin, the barriers between idea and execution are shrinking fast.
For developers, this means faster prototyping and automation of routine tasks.
For content creators, it means more time spent on vision, less on logistics.
For everyone else, it’s a peek at the future of collaborative computing.
Want to Try It Yourself?
You’ll need:
- A WordPress installation with AI Engine plugin
- A Pro Suite hosting account that supports MCP connectivity
- Claude Desktop App with `mcp.js` integration for orchestration
- Optional: MyCelf’s WP AI Agent plugin for visual content generation
With these tools, your AI assistant can become your next web developer.
Further Reading & Resources
Learn the technical specification and use cases for MCP.
Official plugin page, setup instructions, and feature list.
Explore Claude’s capabilities and download the desktop version.
A custom content and image generation tool for WordPress blogs.
View the live AI-built website featured in this case study.
Letters from the Ice* may sound like a poetic fiction, but it’s a practical demonstration of where AI and infrastructure meet. And it’s only the beginning.
