Automattic is currently pushing the boundaries with a structured experiment called Radical Speed Month (RSM). One of the most intriguing results to emerge from this sprint is WooCommerce for Claude. This isn’t a feature announcement or a finalized roadmap commitment; rather, it is a technical experiment designed to see what happens when you give an LLM a deeply structured “brain” specifically tuned for e-commerce operations.
In my 14 years of wrestling with WordPress, I’ve seen countless “AI plugins” that do little more than slap a chat interface on a site. Most of them fail because they treat the store as a flat data source. However, this experiment asks a different question: What does the intelligence layer look like when the AI actually understands the context of a merchant’s business? Consequently, the team built a bridge using the Model Context Protocol (MCP).
Why WooCommerce for Claude is different from simple REST
Most developers assume that connecting AI to Woo just means letting it call REST endpoints. That is a recipe for hallucinations and race conditions. WooCommerce for Claude introduces three specific layers that move beyond simple API calls:
- Analytics Skills: Instead of querying raw
wp_posts, the AI hits pre-aggregated analytics lookup tables. It can answer complex questions about coupon performance or sales drivers with precision. - A Knowledge Layer: This exposes the store profile, catalogue schema, and policies as MCP resources. This means the model has the context of the store before it even attempts a tool call.
- AI-Readiness Scoring: This engine produces a 0–100 score based on product completeness and schema coverage, providing a prioritized list of technical fixes.
Specifically, everything is served through a single endpoint at /wp-json/woocommerce-claude/mcp. There are no background crons or separate sync processes. The plugin only works when an MCP request arrives, which is a pragmatist’s dream for performance. For more on how this protocol works, check out my previous post on WooCommerce AI and secure MCP integrations.
Extending the Intelligence Layer
For extension developers, the most exciting part is the provider pattern. You can register your own extension to feed structured knowledge into the layer the AI sees. This is exactly how we should be building “Agentic” tools—not by hacking core, but by following established patterns.
<?php
/**
* Registering a custom provider for WooCommerce for Claude.
*/
add_action( 'woocommerce_claude_register_providers', function( $registry ) {
// bbioon_ prefix for custom implementation
if ( class_exists( 'My_Extension_Provider' ) ) {
$registry->register( new My_Extension_Provider() );
}
});
I honestly thought I’d seen every way a store could break when attempting AI integration. I’ve seen sites crash because an AI agent tried to loop through 50,000 orders via standard REST. Furthermore, raw data without context leads to the AI suggesting “actions” the merchant can’t actually perform. This “Knowledge Layer” approach fixes that by setting the boundaries upfront. You might want to read about why context is the missing piece in most AI workflows.
Look, if this WooCommerce for Claude stuff is eating up your dev hours, let me handle it. I’ve been wrestling with WordPress since the 4.x days.
How to test it today
If you want to get your hands dirty, the repository is public. You can clone it, run composer install, and activate it on a local environment. I recommend using npx @wordpress/env to boot up a clean instance. The repo even includes a seed script that generates 24 months of realistic order data so you can test the analytics skills properly. Therefore, you don’t have to risk your production database just to see if the AI can actually find your biggest refund leakage.
The post-experiment phase is where the real work begins. Whether this becomes a core feature or stays a separate plugin, the shift toward standardized protocols like MCP is inevitable. If you’re building for the future of commerce, you need to be looking at how Claude and other LLMs interact with your data today.
” queries:[
