The 2026 proposal moves WordPress back to three majors a year, roughly one every four months, with one to three minors in between. Freezing a client site at 6.7 is what taught me why that cadence is easier to live with.
WooCommerce 10.4.3 patches guest order data exposure in the Store API, throttles HPOS sync-on-read during heavy jobs, and restores the cart undo link that 10.4.2 broke.
WordPress itself is free to download, but hosting, premium licenses and maintenance are not. Where the money actually goes once a site has to earn its keep.
WordPress may go back to three major releases in 2026, roughly one every four months. What that faster cadence means for staging, testing and custom code.
Custom REST endpoints leave AI agents guessing. The MCP Adapter v0.3.0 turns registered WordPress abilities into tools an agent can discover, and logs the calls.
A dev on PHP 8.2 white-screened a client host still on 7.4. How standardized local environments, Blueprints and selective sync keep that from happening.
A deleted product used to take down order history with a fatal error. WooCommerce 10.4.2 adds the null check, along with PHP 8.4 transient validation and support for the renamed WordPress 6.9 Abilities API hooks.
A client built a business site in twenty minutes with an AI prototyping tool. It looked great and could not talk to his inventory system or be crawled properly. Here is where AI website building stops and production work starts.
I dismissed WordPress Playground as a browser toy. It now installs almost every popular plugin, runs Composer and Laravel, and spins up a shareable environment per ticket, which has replaced most of my staging work.
Real-time collaboration is targeted at WordPress 7.0 in 2026. Beta feedback says native blocks work fine and legacy post meta does not, so custom fields need registering for the REST API before then.