Vibe Coder

Set up your machine, understand what an agentic CLI is, run your first loop.

1. 01

   What Vibe Coding Means

   core

   You describe outcomes; the agent edits files, runs tests, and commits. You stay on the what; the agent handles the how.

   • lessonWhat Is Claude Code
2. 02

   Terminal Fluency

   core

   Be comfortable in a shell — cd, ls, git, env vars, pipes. Everything happens here.

3. 03

   Install the Agent CLI

   core

   Get Claude Code (and later Codex / Antigravity) installed and authenticated.

   • lessonInstall the CLI
4. 04

   Your First Conversation

   core

   Open a repo, ask for a small change, watch the loop run end-to-end.

   • lessonFirst Conversation

Master the read-edit-test-commit cycle with Claude Code. This is the muscle memory of vibe coding.

1. 01

   The Loop Itself

   core

   Prompt → read → edit → verify → commit. Internalize the rhythm.

   • lessonThe Agentic Loop
2. 02

   CLAUDE.md & Memory

   core

   Project-level instructions that travel with the repo. Make the agent fit your codebase.

   • lessonCLAUDE.md & Memory
3. 03

   Searching & Editing

   core

   How the agent finds and changes code without breaking adjacent context.

   • lessonReading & Editing
   • lessonSearching the Codebase
4. 04

   Tests & Commits

   core

   Verification is what separates vibe coding from vibe-praying. Run tests, commit often.

   • lessonRunning Tests
   • lessonGit & Commits
5. 05

   Debugging with an Agent

   core

   Push back when the agent guesses. Ask for repros, logs, and minimal failing cases.

   • lessonDebugging

Customize the agent with slash commands, subagents, plan mode, hooks, and MCP servers.

1. 01

   Slash Commands

   core

   Reusable shortcuts for the prompts you find yourself typing over and over.

   • lessonSlash Commands
   • lessonCustom Slash Commands
2. 02

   Subagents

   recommended

   Spawn focused workers for searches, reviews, and parallel exploration.

   • lessonSubagents
3. 03

   Plan Mode

   recommended

   Plan-first execution for risky or multi-step changes.

   • lessonPlan Mode
4. 04

   Hooks

   optional

   Automate side-effects on tool events — formatting, linting, notifications.

   • lessonHooks
5. 05

   MCP Servers

   recommended

   Extend the agent's tool surface to your own APIs and services.

   • lessonMCP Servers

Capstones: real features, real PRs, real production releases.

1. 01

   A Next.js Feature, End-to-End

   core

   Spec → implementation → tests → PR. The canonical web-app loop.

   • lessonA Next.js Feature
2. 02

   Tests-First Mode

   recommended

   Write the failing test first; let the agent make it pass.

   • lessonTests-First
3. 03

   Shipping to Production

   core

   Branch protection, CI gates, env management, and confidence to deploy.

   • lessonShipping to Production
4. 04

   Claude Code Capstone

   core

   Build a real, deployable feature from a one-line ask.

   • lessonCapstone

Take the loop to React Native — Expo, navigation, native modules, App Store.

1. 01

   Meet Codex

   core

   What Codex adds when you're targeting mobile and the agent needs a simulator.

   • lessonMeet Codex
2. 02

   Expo Scaffold

   core

   Spin up a real app you can run on your phone in under 10 minutes.

   • lessonScaffold an Expo App
3. 03

   Auth, API, Realtime

   recommended

   Supabase auth, REST data, and realtime channels — the modern mobile stack.

   • lessonSupabase Auth
   • lessonRealtime
4. 04

   App Store Submission

   recommended

   Build, TestFlight, IAP, and getting through review without surprises.

   • lessonTestFlight
   • lessonStore Submission

Antigravity: rosters, plan/build/review/deploy pipelines, human-in-the-loop.

1. 01

   What Antigravity Is

   core

   A multi-agent orchestrator where agents play distinct roles.

   • lessonWhat Is Antigravity
2. 02

   Roster Design

   core

   Pick the smallest set of agents that covers your loop. Don't over-staff.

   • lessonRoster Pattern
3. 03

   Orchestration Patterns

   recommended

   Sequential pipelines, fan-out/fan-in, voting, retries, human-in-the-loop.

   • lessonSequential Pipelines
   • lessonHuman in the Loop
4. 04

   Antigravity Capstone

   core

   A working multi-agent project on a real repo, with budget and observability.

   • lessonCapstone

Habits, communities, and the meta-skill of staying current as the tools shift.

1. 01

   Build Your Prompt Library

   recommended

   Capture the prompts that work. Treat them like a personal codebase.

2. 02

   Context as a Budget

   recommended

   Long sessions decay. Compact, restart, and front-load only what matters.

3. 03

   Stay in the Loop

   optional

   Follow the model release notes; the floor moves every few months.

   • externalAnthropic News↗
   • externalOpenAI Blog↗