claude.ai, Claude Code, and the API are three different tools with three different memory models. Picking the right surface for the right task is a force multiplier.

Two surfaces, two persistence models, one name. How to set up and configure Projects in both claude.ai and Claude Code for efficient, secure, long-running work.

Reusable slash commands that teach Claude how to work. Write once, invoke anywhere — with shell preprocessing, auto-detected project types, and pre-approved tools.

Stepwise design iteration where every stage is a clean, self-contained project. How to clone, clean, and build forward without losing any prior state.

Tokens are the unit of everything your AI integration does. What they are, where they accumulate invisibly, and how to stop rate limits from becoming a production reliability problem.

The live preview panel in claude.ai that renders HTML, React, and SVG outputs immediately. How to iterate on them in conversation, what libraries are available, and how to go from prototype to deployed tool.

The open protocol that gives AI models a standard way to reach tools, data, and services. What it is, how to write a server, and the security tradeoffs to understand before you deploy one.

Programmatic access to Claude — what it is, what it costs, how token billing differs from a subscription, and a working Python app from zero to first call.