A practical guide to working effectively with AI coding assistants, covering workflow design with explicit start/stop words, persistent rules over per-session instructions, breaking tasks into atomic units, feeding good code examples as context, critically reviewing AI-generated tests, knowing when to interrupt or kill a session, and never letting the assistant commit without explicit approval. The core thesis: AI assistants are force multipliers, not autonomous problem-solvers, and structured prompting discipline is what separates productive use from chaos.
Table of contents
TypographyWorkflowsContext Is King (And Your Assistant Has Amnesia)Atomic Tasks, or: Stop Writing NovelsFeed It Good Code, Get Good CodeTrust, But Verify (Especially the Tests)The Art of Saying “No, Try Again”Know When to Kill the SessionThe Checklist434 Impressions