Multi-Agent Coding: Orchestration Patterns From Production
What multi-agent coding actually looks like in production: pipelines, adversarial review, fan-out, and the failure modes that only show up when it runs daily.
Writing
Notes on building with AI — development, agents, and creativity.
Agentic Coding: What Actually Changes About the Job
Agentic coding doesn't just speed up programming — it moves the job. Specification becomes the work, review becomes the bottleneck, judgment stays yours.
Claude Code Tips: Seven Habits Worth Building Early
Seven small habits that make Claude Code more useful: ask for the plan before the diff, point at real files, and hand over the command that proves the work.
Context Engineering in Practice: a Working System
What context engineering actually is, the five layers of a context system that runs a real portfolio daily, and how to build yours — from the first file to unattended agents.
Claude Code vs Cursor: a Builder's Honest Take
Claude Code vs Cursor from someone whose companies run on one of them daily — where each tool genuinely wins, and the one question that decides which you need.
Claude Code Skills: the Complete Working Guide
What Claude Code skills are, when they fire, and how to build ones that actually get used — from someone running eleven production repos on them daily.
Standardizing eleven repos in a day (with an AI chief of staff)
What happened when I handed my entire codebase portfolio to an AI agent with a two-phase brief: audit everything, then — with my approval — fix everything.