When my LLM coding assistant hit a wall on tasks, I had it decompile a reference app from the Play Store. analyzed the flow and saved me a week of research.
The LLM era moves faster than any of us can follow. Here's proof — and a quiet place to start.
A developer deep in small-model experiments realizes the most important part of his stack — the space between the harness and the model — is the only part he can't watch without leaving his workflow. So he builds a transparent proxy to put eyes on the middle.
A manifesto for professional AI-assisted engineering. Why 'prompting' fails and 'architecting' survives — and how the chain from research to plan to implementation keeps human intent sovereign while the machine does the typing.
Why explicit, boring, low-magic code keeps working when AI agents get small — and what 'magic' stacks quietly cost you.
A cross-lingual word chain game born from cultural humor. three languages, and one very confused AI.
How two native OpenCode tools let agents analyze their own conversation history — and what I accidentally discovered along the way.
How I built a simple AI agent to transform my coding explorations into blog posts—without becoming a writer.