Gemini CLI

Start Gemini CLI in a practice project and ask it to explain the files before changing anything. You will finish with an evidence-based project map and a small next step you understand.

Before you start

• A terminal, internet connection, a browser for sign-in, and Node.js 20 or later with npm. Google recommends macOS 15+, Windows 11 24H2+, or Ubuntu 20.04+.
• A Google Account eligible for Gemini CLI in your location. This walkthrough uses Google sign-in; some organization and Code Assist accounts also require a configured Google Cloud project.
• A practice copy of a small project you own, including its README if it has one. Remove secrets and customer data from that copy before using it.
• A little terminal familiarity: you should be able to open a terminal and move into a folder with cd.

Want to do it yourself?

Difficulty

6 out of 10

Estimated completion time

30–60 minutes

Skills you’ll need: Check your runtime, install a CLI, navigate a practice folder, and compare its project explanation with actual files.

I’ll do it myselfI’d rather get help

How this helps your brand

A verified project map helps you understand where a future website or software change belongs. Checking real files and documented commands gives the next task a more dependable starting point.

1. 1

   Step 1: Check the runtime and install the CLI

   In your terminal, run node --version and npm --version. If Node is below version 20 or either command is missing, set up a compatible Node.js installation before continuing.

   Run npm install -g @google/gemini-cli to install the default stable release. If gemini later reports “command not found,” check whether npm’s global binary directory is in your PATH. Google’s installation guide also documents npx @google/gemini-cli as an alternative launch method.

2. 2

   Step 2: Open the practice folder and sign in

   Use cd followed by your practice folder’s path, putting a path containing spaces in quotes. Launch Gemini CLI by running gemini. Select Sign in with Google and complete the browser sign-in, then return to the terminal.

   If your account requires a Google Cloud project, follow the official authentication setup with your administrator. Do not paste a password or API key into the conversation to resolve a sign-in error.

   Illustrated exampleFrom your shell to the Gemini CLI conversation

   Example terminal session

   1 Shell · before chat

   $cd "Northline practice"

   Current folder: Northline practice/

   $gemini

   2

   Sign in with GoogleComplete browser sign-in → return here

   3 Gemini CLI · conversation

   Explain this project and cite local file paths. Do not edit files or run the application.

   Explanation only

   1. 1
      Set the folder first

      The CLI starts inside your practice copy. Quotes keep a folder name with spaces together.

   2. 2
      Use the browser for sign-in

      Return to the terminal when the account is connected.

   3. 3
      Then describe the task

      Shell commands launch the tool. Your project question goes in the CLI conversation.

3. 3

   Step 3: Ask it to map the project

   Paste the prompt below into the Gemini CLI input, using the project’s name. Ask it to inspect local files and identify the entry point, important folders, and documented run commands.

   Keep this first task to explanation. If the CLI asks to run a command or write a file, read the proposed action. Decline edits or installations that are unnecessary for understanding the project.

4. 4

   Step 4: Check the explanation against actual files

   Open two or three files the answer names. Check that they exist and that their contents support the explanation. Compare suggested run commands with the project’s README and configuration before trying them.

   If Gemini names a file that does not exist, give it the correct path and ask it to revise the map using that file. An absent README is a reason to inspect the project configuration, not invent a startup command.

   Illustrated exampleMatch the project map to file evidence

   Northline practice/

   • README.md1
   • package.json1
   • src/
     • main.tsx2
     • App.tsx3

   Open the files named in the answer.

   1 Documented run command

   README.md says npm run dev.

   package.json agrees: "dev": "vite".

   ✓ Both files support the command

   2 App entry point

   src/main.tsx mounts the app:

   createRoot(root).render(<App />)

   3 Home-page heading

   src/App.tsx contains the visible text:

   <h1>Northline Studio</h1>

   ?

   No test script in this example? Record “Test command not established” and ask what evidence is missing.

5. 5

   Step 5: Choose one next step without changing the project

   Ask which one small change would help you learn the project, and what files and checks it would involve. For a simple website, that might be changing one heading and checking it in a local preview.

   Save the verified project map and proposed next step in your own notes. If a quota or access error interrupts the session, keep the error text and consult the CLI’s account and troubleshooting documentation before retrying.

Help me understand this local project: [project name]. Inspect its README, configuration, and relevant source files. Explain the purpose, entry point, main folders, and any documented commands for running or checking it. Cite the local file paths behind your explanation.

This task is explanation only: do not edit files, install dependencies, start servers, or run the application. If information is missing, say what you could not establish. Finish by proposing one small beginner-friendly change and how I could verify it, but do not implement it.