Plan Mode: Make Claude Code Think Before It Touches Anything
The fastest way to lose an afternoon is letting an AI agent start editing on a vague task. Plan mode forces it to research and propose first — relocating review to where it's cheap.
#ai agents
8 articles
The Context Window Is Your Real Constraint
When an AI session starts forgetting and contradicting itself, it's usually not intelligence that ran out — it's context. How to manage the AI's working memory like the scarce resource it is.
Never Trust Your AI's Confident Answer
An AI assistant is exactly as fluent and authoritative when it's wrong as when it's right. That symmetry is the trap — and calibration is the skill that beats it.
Using Claude Code for Things That Aren't Code
The name is a trap. Strip away the framing and it's a general reasoning agent that can read, reason, and act — code is just the most common thing people point it at.
Parallelizing Work with Claude Code Subagents
One AI session has a context ceiling. Subagents let Claude Code fan work out to isolated helpers — and report back only what matters.
Writing a CLAUDE.md That Actually Changes Claude Code's Behavior
A CLAUDE.md that gets ignored is worse than none. How the file loads, the hierarchy that decides which rules win, and why specificity is everything.
Claude Code Hooks: Turning Requests Into Rules
Stop asking Claude Code to format your code, run your tests, or avoid your migrations. Hooks let the harness enforce it — deterministically, every time.
I Gave Claude Code Long-Term Memory
AI coding assistants forget everything between sessions. Here's the persistent-memory system I built so mine finally remembers my projects, my decisions, and my mistakes.