Vibe Coding for Beginners: From Zero to App/Course resources
Lesson 1 / 2 · 10 min · 0 / 3 exercises
Course contents
Concept · 3 hands-on exercises

Your vibe coding cheat sheet

Keep the prompts, commands and terms you'll use most in one place, for every tool in the course.

This page collects the prompts, commands and terms from the course in one place, so bookmark it. Everything here was checked against each tool's documentation in November 2026. Commands change, so if one stops working, ask your agent “what's the current command for this?” and confirm it in the tool's docs.

Prompts worth reusing

When you want to…Say
Start something new“Before writing code, ask me up to five questions. Then show your plan and wait for my OK.”
Make one change“On [screen or page], [what's wrong]. Change only that, and don't touch anything else.”
Report a bug“When I [steps], I expected [X] but got [Y]. Console: [exact error]. Explain the cause first, then fix it.”
Understand a change“Explain what you changed and why, in plain English, file by file.”
Check the agent's work“Show me how you know this works. What exactly did you test?”
Review security“Do a security review of this project. Don't change anything yet. List each finding with the file, severity and a one-sentence fix.”
Finish a task“Commit this with a message that says what changed and why.”

Commands by tool, as of November 2026

TaskClaude CodeCodexCursorGrok Build
Start in a projectclaudecodexOpen the folder, then Cmd+I or Ctrl+Igrok
Plan before buildingShift+Tab to plan mode, or /planAsk for a plan in the promptShift+Tab to Plan ModeAsk for a plan in the prompt
Change what runs without askingShift+Tab cycles the modes/permissionsRun mode in Agent settingsSee its Modes and Commands docs
Create an instruction file/init (writes CLAUDE.md)/init (writes AGENTS.md)/create-rule, or an AGENTS.mdAsk for an AGENTS.md; grok inspect lists what it found
Start fresh on a new task/clearStart a new sessionStart a new chatStart a new session
Undo the agent's changes/rewind, or Esc twiceRestore your last commitRestore CheckpointRestore your last commit
Review before shipping/security-review (needs an origin remote)/reviewReview the diffReview the diff

Google's terminal agent for individuals is Antigravity CLI, which starts with agy. Gemini CLI, now aimed at Gemini Code Assist licenses, starts with gemini. In every tool, git is the undo that always works: “discard the uncommitted changes and go back to the last commit.”

Browser checks you'll use constantly

  • View the page source: right-click the page and choose View Page Source. Search it with Cmd+F or Ctrl+F.
  • Open the console: right-click, choose Inspect, then the Console tab. Red lines are errors.
  • Preview phone sizes: in Chrome or Edge, Inspect, then the device toolbar (the phone-and-tablet icon). Firefox calls it Responsive Design Mode.
  • See what the page sends: Inspect, then the Network tab, then use the page.
  • Test as a stranger: open a private window, signed out of everything.
Quick checkExercise 1 of 3

In Claude Code, which command gives you a clean conversation for a new task while keeping project memory?

Words you'll meet

Match them upExercise 2 of 3

Match each term to its meaning.

Pick one on the left, then its partner on the right.

Term
Meaning
0 of 8 matched
Fill the blanksExercise 3 of 3

Complete the bug-report template so the agent has what it needs.

When I , I expected the new item to appear, but the section went . Console: SyntaxError at line 9. Explain the first, then fix it.
Check your understanding
Q1. Which prompt helps most when a change surprises you?
Q2. What does an environment variable protect?
· Work through the 3 remaining exercises above (stuck? each one can show you the answer).
· Score 100% on the quiz.
Vocabulary in this lesson
AgentA model wired to tools and a loop: it reads, acts, checks the result, and goes again until the task is done or you stop it.AGENTS.mdThe instructions file at the repo root that every session reads first: commands, conventions, gotchas. CLAUDE.md in Claude Code.ToolA named capability the harness lets the model invoke, such as reading a file, running a shell command or fetching a page.Plan modeA read-only agent mode: the agent may search and read, but must propose a plan for your approval before it edits anything.ClearingWiping the session so the next request starts from an empty context window, keeping only the standing instructions.Context windowThe fixed-size buffer of tokens a model can read in one request; if something isn't in it, the model can't see it.

Plain-English definitions from the AI Coding Dictionary.