THE CONCEPT LIBRARY
One idea at a time.
AI basics you can understand and use. With a short, an example and a small exercise.
-
▶ 60 SEC
Context & memory
Auto Memory in Claude Code: Context Across Sessions
Learn what Claude Code Auto Memory stores, how it differs from CLAUDE.md, and why important project rules still need a human-maintained home.
-
▶ 60 SEC
Better prompts
Caveman: Shorter Agent Replies Aren’t Token Savings
Learn what the Caveman response-style skill changes, what it preserves, and why concise output does not guarantee context compression or lower costs.
-
▶ 60 SEC
Context & memory
CLAUDE.md: Shared Instructions, Not a Security Lock
Learn how a small, versioned CLAUDE.md gives Claude Code durable project context—and when local files, Skills, or Hooks are a better fit.
-
▶ 60 SEC
Tools
Ponytail: Check First, Then Ask Your Coding Agent for Less Code
Ponytail is a practical decision ladder for checking what your project and platform already provide before an AI coding agent adds more code.
-
▶ 60 SEC
Better prompts
Prompting: Set effort, scope, and completion clearly
A beginner pattern for making AI coding requests more predictable: state the goal, boundaries, question rule, and proof of completion.
-
▶ 60 SEC
Context & memory
Token optimization: give AI coding tasks the right context
Learn to measure AI request size, keep essential project rules available, and load reference material only when it is relevant.
-
▶ 60 SEC
Better prompts
Write a clear AI task: goal, context, result
A simple prompt structure helps you ask coding assistants for answers you can inspect and reuse.