The 2026 shift from prompting an agent to writing the loop it runs inside. What loop engineering is, the building blocks, and the four non-negotiables that make a self-correcting agent loop trustworthy.

How Anthropic's Claude Code inspired me to build a multi-agent coordination plugin for pi, and why one AI agent is no longer enough.

I built three pi extensions that work together to create the most productive AI coding workflow I've ever used. Here's how plan mode, worktrees, and multi-agent teams combine into something greater than the sum of their parts.

Boris Cherny ships 10-30 PRs a day using Claude Code. Here's exactly how he does it, and how you can too.