One of the biggest modern creations is AI, and it has changed the way we do everything. One of the bigger places where AI has taken over is coding. Today, you’ll find tons of AI coding assistants for you to pick and choose from.
And now, there’s a new player in town to take over all AI coding assistants: Claude Code. So, if you are a seasoned coder or a newbie in coding, using an assistant to help you with code writing is important. And today, we will be taking a look at everything you need to know about Claude Code. Let’s get started.
What is Claude Code?
Claude Code is an AI coding assistant that has been developed by Anthropic. This AI coding assistant can be integrated directly into IDEs, workflows, and terminals that can help coders and developers edit and build codebases with prompts in simple natural languages. Claude Code is powered by Claude 3.7 Sonnet: an advanced AI model that uses hybrid reasoning and extended thinking mode to complete various tasks, including coding.
Setting up Claude Code
Let’s now take a look at what the system requirements are and what systems can install and use Claude Code, as well as other things that you will need.
System Requirements for Claude Code
Check out the system requirements that you need to have in order to use Claude Code.
- RAM: 8 GB
- Storage Space: 5 GB or more
- CPU: Any CPU that has support for virtualization (necessary for WSL2 on Windows)
- Operating System: Windows 10 with WSL 2, macOS 10.15, Ubuntu 20.04, Debian 10+
Prerequisites for Claude Code
Having a PC that meets all the requirements is not a big task. There are certain things that you must have. Without these things, you cannot use Claude Code at all.
Claude Plans and Pricing
Now, to use Claude Code, you will need to get one of the paid plans. The free plan from Claude will not work with Claude Code. Here is a list of the paid plans, prices, and their benefits.
| Claude Plan | Price | Usage Capacity | Use Case |
| Claude Pro | $20 per month or $200 annually | Standards et of limits | Good for regular and day-to-day users |
| Claude Max 5x | $100 per month | 5x Pro capacity per session | Best for users who depend on Claude for work and various tasks. |
| Max 20x | $200 per month | 20x the capacity of the Claud Pro plan | Best for users who run Claude at scale. Also, best for teams for easy collaborations. |
Installing Claude Code
Once you have an active plan from Claude, let’s take a look at the different ways to install Claude Code on your PC.
Installing Claude Code with Local LLMs
Now, if you are looking for a simple way to try to use Claude Code on your PC, in this method, we will be looking at how you can use Claude Code with Ollama. Let’s get started.
- Make a folder of your choice in a location that you can remember. This folder will be used to store, access, and edit your projects with Claude Code.
- Open the folder and in the address bar, type in cmd to open Command Prompt.
- Open the web browser and head over to the Ollama website.
- Now, click on the download button to download Ollama to your PC.
- Choose the LLM of your choice that you want to install locally on your PC.
- The Models tab on the Ollama website has several models, so pick the one that you want to pick.
- Once you have installed the model, type ollama in Command Prompt and hit the enter key.
- It will show you the models that are installed on your machine. Use the arrow keys to select your model.
- In my case, I will be selecting Claude Code with minimax 2.5 cloud
- Once you select the Enter Key, you will get Claude Code running right away in Command Prompt, Windows PowerShell, or whatever Terminal you selected in the first place.
On the very first screen, you will get to choose the theme of your choice. Use the arrow keys to select the theme. Now, you will see a small message saying that Claude can make mistakes and that it is necessary to review Claude’s code and responses.
Claude Code will ask you if you wish to trust the folder. This is the same folder where you run Command Prompt from. If you know the folder, select the Yes, I trust this folder option. Claude Code will now be running in Command Prompt right away.
Now, you can write your prompts and codes within Terminal or Command Prompt right away. But did you know that you can also connect Claude Code with the IDE of your choice?
Connect Claude Code with VS Code
Here’s what you will need to do to connect Claude Code with VS Code
- Make sure that you have the latest version of VS Code installed. If not, download and install it from here.
- Download the extension from this link for VS Code.
- Or, you can open the Extensions view in VS Code(Cmd+Shift+X for macOS or Ctrl+Shift+X for Windows and Linux devices) and search Claude Code to install the extension.
- Now, you should see the Claude Code icon within VS Code.
- Open up the Claude Code panel in VS Code by pressing the Cmd+Shift+P for macOS or Ctrl+Shift+P for Windows and Linux devices.
- And that is it. You can choose to follow the onboarding process for Claude Code or simply start away by using prompts to code with Claude Code in VS Code.
Connecting Apps with Claude
There are two ways you can select to connect your favorite apps and services to Claude. You can pick the API Key way or the OAuth method. Let’s take a look.
Connecting Apps via 0Auth to Claude
This is a simple and easy-to-use method to connect your favorite apps and services. Here is how you can connect Claude to your apps or services
- Launch the Claude desktop app and click on the profile icon in the bottom left corner.
- Now, select Settings.
- Choose Connectors followed by Browse Connectors
- Search for the app or service that you wish to connect to and click on the little + icon.
- The app or service’s login page will open in your web browser. Sign in with the respective account.
- With the sign-in completed, go back to Claude’s Connectors page. You will see the connected app or service displayed right away.
Connecting Apps via API Key to Claude
While it is easy to connect apps via 0Auth in the Claude app, there are a few disadvantages:
- Limited number of apps
- Certain services and apps only give read-only access
- No option to customize how you want the app to connect with Claude
Because of these reasons, many users prefer to make use of Custom Connectors with API Keys with Claude. This lets you connect any app of your choice that is not present in the list, gives you access to read and write data with the app or service, and you get to control everything between the app and Claude.
However, you need to make sure that you are not connecting Claude with apps, services, or systems that store very important data. You do not want Claude to modify or delete this type of data. So make sure that you know what you are connecting with Claude, and then connect the app or sieve to it.
- Visit the Anthropic Console by clicking this link.
- Now, create your account or log into it if you already have one.
- Click on API Keys and select Create Key.
- Copy this API key and save it in a secure location.
- Now, install the SDK using either Node.js or via Python by using these commands
- Node.js: npm install @anthropic-ai/sdk
- Python: pip install anthropic
- Next, you need to save the acquired API key as an environment variable.
- Windows OS: setx ANTHROPIC_API_KEY “insert API key here”
- macOS/Linux: export ANTHROPIC_API_KEY=”insert API key here”
- With the API key setup, it’s time to enter the key into the backend of the app or service using Node.js or Python. Once you have it set up, simply connect it to the backend of your app or service.
Node.js example:
import Anthropic from "@anthropic-ai/sdk";
const client = new Anthropic({
apiKey: process.env.ANTHROPIC_API_KEY,
});
async function run() {
const response = await client.messages.create({
model: "claude-3-5-sonnet-20241022",
max_tokens: 500,
messages: [
{ role: "user", content: "Hello Claude" }
],
});
console.log(response.content[0].text);
}
run();Python Example:
import anthropic
import os
client = anthropic.Anthropic(
api_key=os.environ["ANTHROPIC_API_KEY"]
)
response = client.messages.create(
model="claude-3-5-sonnet-20241022",
max_tokens=500,
messages=[
{"role": "user", "content": "Hello Claude"}
]
)
print(response.content[0].text)Important Points to Remember
- Never call Claude from the front end.
- Keep a log of all errors
- Avoid using the wrong model names. Always double check.
- Never skip using backend proxies.
Starting Your First Session
Once you have set up Claude Code with the model of your choice, you need to create a project directory. You can do this by creating a folder in a specific location and then running Command Prompt, Terminal or Windows PowerShell from the folder. This fodder becomes your project directory when using Claude Code.
Now, you can ask Claude Code, and it will provide you with the necessary responses. With the necessary permissions granted, Claude Code will now have access to the following:
- Apps and services that are connected via extensions.
- Claude.md file
- Desktop IDEs
- Git State
- Project Directory
- Your Terminal, Command Prompt or Windows PowerShell
Claude.md File: What is It?
A Claude.md file is a project markdown file that is used by Claude Code to read for various instructions that have been set up by the user. These are strict instructions that Claude Code will follow and prioritize over the basic prompts that are provided by the user in the Terminal or IDE. You can create the Claude.md file manually and place it in your project’s directory. Claude Code will automatically detect this file and follow all instructions present in the file.
What Should Your Claude.md Contain?
There are a number of elements that need to be present in the Claude.md file so that Claude can understand what it needs to do. Here’s the checklist of things that have to be present in the file
- Define the role of what Claude is going to do in the project
- Pointers on what it needs to consider as a good output
- Instructions on how to put the responses in a particular style, tone, and formatting.
- Providing domain context so that Claude knows how to base these responses requested by the user.
- Defining the capabilities and tasks that Claude needs to perform.
- A set of safety rules or restrictions that Claude should follow
- Set instructions if Claude needs to use connected apps or services
- Rules on how to handle errors
- Policies on how to provide clarifications
While these are important elements that need to be added to your Claude.md file, there are a lot of things that you should avoid entering on your Claude.md file.
- Ever-changing data.
- Instructions that cancel out each other- say no to redundant instructions
- Long paragraphs and essays- keep it short and simple
- Unnecessary marketing style fluff
Here is a sample of what your Claude.md file should look like. It doesn’t have to be the same, but as long as you keep the core elements in your file, Claude will follow them strictly.
# Claude Assistant Specification
## Overview
You are an AI assistant embedded inside a web application.
Your role is to help users complete tasks accurately, safely, and efficiently.
You must follow the behavioral and technical rules defined in this document at all times.
—
## Core Objectives
1. Provide correct, actionable answers.
2. Be concise but complete.
3. Ask clarifying questions when requirements are unclear.
4. Never fabricate APIs, data, or capabilities.
5. Prefer structured responses when useful.
—
## Tone & Communication Style
– Professional and direct
– Minimal fluff
– Clear step-by-step instructions for technical tasks
– Use bullet points and code blocks when appropriate
– Avoid emojis unless explicitly requested
—
## Domain Context
This assistant operates inside a **developer web app**.
Users will commonly ask about:
– APIs
– Backend architecture
– Databases
– Debugging
– SDK usage
– Integrations
Assume the user is technical unless stated otherwise.
—
## Allowed Capabilities
You may:
– Explain concepts
– Generate code
– Refactor code
– Suggest architecture
– Draft API calls
– Troubleshoot errors
– Produce JSON outputs
You must:
– Prefer practical solutions
– Include examples
– Show full working snippets when writing code
—
## Tool Usage Rules
If tools are available, follow:
### When to call tools
Call a tool if:
– The user asks for real-time data
– The task requires external execution
– The task requires database access
– The task involves file operations
### When NOT to call tools
Do not call tools if:
– You can answer from reasoning alone
– The user asks for conceptual help
– The request is hypothetical
—
## Response Formatting Rules
### Code
– Always wrap code in fenced blocks
– Include language tags
– Provide runnable examples
– Avoid pseudocode unless requested
### Step-by-step tasks
Use numbered lists.
### JSON outputs
If user asks for JSON:
– Return valid JSON only
– No extra commentary
—
## Error Handling Behavior
If the user provides broken code:
1. Identify the issue
2. Explain why it fails
3. Provide corrected version
4. Suggest improvements
—
## Clarification Policy
Ask questions if:
– Requirements are ambiguous
– Multiple approaches exist
– Security implications are unclear
Do NOT assume missing details in critical tasks.
—
## Security Rules
Never:
– Expose API keys
– Suggest insecure practices
– Bypass authentication
– Generate malware
Always:
– Recommend environment variables
– Suggest secure defaults
—
## Example Interaction
**User:**
How do I call Claude API from Node?
**Assistant should:**
– Provide install steps
– Show minimal working example
– Use official SDK
– Explain env setup
—
## Default Assumptions
Unless stated otherwise:
– Backend: Node.js
– Frontend: React
– API: REST
– Auth: Bearer tokens
– Environment: Production-capable
—
## Response Length
Default: concise
If user says “explain deeply” → expand
—
## Priority Order
If conflicts occur, follow this order:
1. Safety rules
2. Tool rules
3. User request
4. Brevity
Keyboard Shortcuts for Claude Code
Here is a list of shortcuts that you need to keep in mind when using Claude Code
Prompt/Execution Shortcuts
- Ctrl + C: Stop generation (terminal)
- Ctrl + Enter: Force send (in some environments)
- Ctrl + L: Clear screen / conversation (CLI)
- Esc: Cancel current generation
- Shift + Enter: New line without sending
Conversation Navigation Shortcuts
- Ctrl + Down Arrow: Next prompt in history
- Ctrl + R: Search command history (terminal)
- Ctrl + Up Arrow: Previous prompt in history
- Page Down: Scroll down
- Page Up: Scroll up conversation
- Up Arrow / Down Arrow: Scroll through history
Shortcuts for Editing Inside the Prompt Box
- Alt + left Arrow: Move back one word
- Alt + Right Arrow: Move forward one word
- Ctrl + A: Move to start of line
- Ctrl + E: Move to end of line
- Ctrl + K: Delete to end of line
- Ctrl + U: Delete entire line
- Ctrl + W: Delete previous word
Code Block Output Shortcuts
- Cmd/Ctrl + Click: Open file reference
- Ctrl + Shift + C : Copy code block
- Ctrl + Shift + Enter: Run generated command (CLI environments only)
- Tab: Accept suggestion (specific editors only)
Project and File Shortcuts
- Ctrl + “: Toggle terminal
- Ctrl + B: Toggle sidebar
- Ctrl + F: Find
- Ctrl + P: Open file search
- Ctrl + S: Save file
- Ctrl + Shift + F: Find in project
- Ctrl + Shift + P: Command palette
Agent Control Shortcuts
- Ctrl + C: Hard stop task
- Ctrl + D: Exit session
- Ctrl + Z: Background process (terminal)
- Esc: Interrupt Claude
Important Commands in Claude
- /clear
- /exit
- /files
- /help
- /reset
- /tools
Now, different shortcuts work differently based on the environments that you are using Claude Code such as VS Code, Terminal, Cursor and other web based developer tools.
Closing Thoughts
This concludes everything that you need to know about Claude Code and how you can set it up and implement it in your coding workflow. What do you think about Claude Code? Is this an AI assistant that you would like to use with your workflows? Share your thoughts in the comments section.
Explore More:
