Latest field notes
6 piecesPrompt-to-Plugin: shipping a working WordPress plugin from a single spec
One plain-text spec, an LLM scaffold, three non-negotiable gates — a pipeline that ships faster without shipping weaker.
DebuggingWhere AI-generated WordPress code quietly breaks — and how to catch it
The LLM writes clean-looking code until it doesn't. These are the five failure modes that pass code review and fail in production.
BlocksVibe-coding a Gutenberg block with an LLM in the loop
Blocks are where LLMs drift hardest — edit/save mismatches, missing block.json, hardcoded markup. Here's the loop that keeps them honest.
SecurityThe security cost of AI-written plugins: the gaps models skip
Nonces, capability checks, escaping, prepared statements — the four guards LLMs omit because the happy path works without them.
AI BehaviorHidden State Drift in your dev loop: why the model forgets your codebase mid-session
Three hours in, the AI reintroduces the meta boxes you deleted an hour ago. That failure mode has a name — and countermeasures.
WorkflowFrom chat to committed: a repeatable AI WordPress workflow
A five-stage pipeline that turns a chat window into shippable WordPress — with the review and test gates that make it safe to move fast.