About this blog

Coding agents are now first-class users of software. They read documentation, inspect repositories, run CLIs, call APIs, change configuration, recover from failures, and ship systems — often with no human watching the terminal. Agent-native programmingis the practice of designing interfaces so that the correct action is obvious to a program, not just discoverable by a person.

That means predictable and discoverable surfaces, typed and structured output, commands that are non-interactive when asked, side effects that are explicit, operations that are idempotent where possible, and errors that actually help you recover. Interfaces that are merelyusable by a careful human are frequently ambiguous to an agent. We write about closing that gap.

Run402 is a platform for exactly these users — agents that build and deploy real backends. We publish this series because the principles are worth writing down one at a time, and because some articles will eventually ship an open-source skill that teaches a coding agent to apply the principle directly.

The blog is open. Every article is Markdown or MDX committed to Git — there is no CMS and no database. The site itself is a static Astro builddeployed on Run402. Dogfooding is the point: the friction of deploying and maintaining a real site on our own platform should become visible to us before it reaches anyone else.