WordPress just introduced the Plugin Directory MCP Server, and if you are living in AI-powered IDEs like Cursor, Claude, or VS Code, your workflow is about to change. For a long time, the bridge between our local dev environment and the official WordPress.org repository has been a manual slog of web forms and readme checks. This new implementation of the Model Context Protocol (MCP) changes that by giving your AI assistant direct context of the directory’s rules and status.
What is the Plugin Directory MCP Server?
In simple terms, the Plugin Directory MCP Server acts as a standardized interface. It allows your AI model to “understand” the current state of your plugin in the context of the official repository. Specifically, it provides tools for readme validation, checking your submission status, and even submitting plugins for review. Furthermore, it gives the AI access to the latest plugin guidelines so it can advise you on compliance before you ship.
I’ve written before about how to simplify AI WordPress plugin development, but this official server takes things to the infrastructure level. Instead of pasting guidelines into a chat window, the assistant “knows” them because it’s connected to the source.
Key Features for Developers
- Readme Validation: Automatically check if your
readme.txtmeets the requirements before you upload. - Submission Status: Query the server to see where your plugin stands in the review queue without refreshing the meta-site.
- Guideline Context: The AI can cross-reference your code against the official WordPress Plugin Guidelines in real-time.
To use this, you typically need to add the server URL to your MCP-compatible tool. In Cursor, for example, you’d add the endpoint in your project settings to allow the model to call these specific “tools.”
{
"mcpServers": {
"wp-plugin-directory": {
"command": "npx",
"args": ["-y", "@wordpress/mcp-server-plugin-directory"]
}
}
}The Senior Dev Reality Check
However, don’t let the automation fool you into laziness. WordPress is very clear on one point: You are responsible for the code. The plugin review team doesn’t care if an AI wrote a race condition or a security vulnerability; the guidelines apply regardless of the author. Consequently, you must refactor and debug everything your assistant generates. This tool is a bottleneck remover for the process, not a replacement for your brain.
I’ve seen too many “AI-first” plugins get rejected because the developer didn’t realize the model was hallucinating a legacy hook that was deprecated in 5.8. Use the Model Context Protocol to verify your assumptions, not to skip the work.
Look, if this Plugin Directory MCP Server stuff is eating up your dev hours, let me handle it. I’ve been wrestling with WordPress since the 4.x days.
Next Steps
This is a major step toward “Agentic Commerce” and more automated workflows in the ecosystem. To get started, check out the official announcement on Make WordPress. If you are serious about modernizing your plugin development stack, integrating this server is no longer optional—it’s a massive productivity hack.
