Claude Code as a Vibe-Coder's Terminal Workshop

Claude Code is Anthropic's terminal-based coding agent. You type claude in a project folder and a chat opens right in your terminal. It reads files, writes code, runs commands, and — most importantly — keeps working on a task for a long time without losing the plot.

• It handles multi-step tasks — build this feature end to end
• It respects a CLAUDE.md file where you write project rules
• It can run tests, start dev servers, and read errors without you copy-pasting
• It shows a plan before doing big changes

# Project: PupLoop Landing ## What this is A marketing site for a dog-walking service. Next.js, Tailwind, shadcn. ## Rules - Do not add new dependencies without asking. - Keep all copy under 140 characters. - Brand color is soft green (#6EBF8B). - Prefer shadcn components over custom markup. ## How I work I am a non-coder. Explain trade-offs in plain English. Before big changes, show me a plan. After each change, run `pnpm build` and report.

Pick something too big for one prompt — a feature with three or four parts. Describe the whole thing at once and tell Claude Code to plan before coding. Let it drive for 20 minutes and check back.

Add a blog to this site. - MDX files in /content/blog - Index page at /blog with cards in reverse date order - Single post page at /blog/[slug] - RSS feed at /rss.xml - Use the existing layout and typography Make a plan first. Then implement one file at a time and run pnpm build after each step.

The big idea: Claude Code is the most patient coding partner you can buy. Give it context, let it plan, and it will happily do an afternoon of work while you watch and steer.