"You will never learn anything new [if it's not] significantly different from what you already know"
Adapted from Rich Hickey - Simple made easy
Spec-driven development is useful when the spec is not a stale document. I use short-lived living plan files as temporary execution contracts for one bounded feature or change so discovery, implementation, checks, docs, and implementation discoveries stay connected while the work is active.
Learn how docs-first product development can make AI coding agents safer by documenting product meaning first, splitting global and co-located docs, and giving future sessions a stable source of truth.
How to Succeed with Agentic Engineering - Part 3
Learn how to make a codebase easier for AI coding agents to navigate and validate using repo-local documentation, behavior specs, executable architectural guardrails, fitness functions, and narrow verification loops.
How to Succeed with Agentic Engineering - Part 2
Learn how to design a maintainable, AI-agent-friendly codebase. The key principles are locality, small blast radius, boundary integrity, navigability, cohesive modules, ownership-aligned boundaries, and restrained complexity, so coding agents can build a correct local model and change software safely.