DEV Community: Md. Shafiuddin Shajib

Use Free Claude Code with Opus 5 in VS Code: CLI and Extension Setup

Md. Shafiuddin Shajib — Fri, 21 Aug 2026 08:17:08 +0000

If you want to get your first practical experience with Claude Code and Opus 5, this setup is worth trying while free promotional credits are available.

The focus is simple:

Claude Code + Opus 5 + VS Code + CLI + Extension

Get the Free AI Credits

Start here:

👉 Get Free AI Credits for Claude Code

Sign up with GitHub and complete the required verification. If your account is eligible, check the available credits in your dashboard and create an API key.

Keep your API key private. Never commit it to GitHub or share it publicly.

1. Install Node.js

Claude Code requires Node.js 18 or later.

Check your version:

node --version

2. Install Claude Code CLI

Open PowerShell or the VS Code terminal:

npm install -g @anthropic-ai/claude-code@latest

Verify the installation:

claude --version

3. Configure Claude Code

On Windows PowerShell, set the required environment variables:

[Environment]::SetEnvironmentVariable(
  "ANTHROPIC_AUTH_TOKEN",
  "YOUR_AGENTROUTER_API_KEY",
  "User"
)

[Environment]::SetEnvironmentVariable(
  "ANTHROPIC_BASE_URL",
  "https://agentrouter.org",
  "User"
)

[Environment]::SetEnvironmentVariable(
  "ANTHROPIC_MODEL",
  "claude-opus-5",
  "User"
)

Replace YOUR_AGENTROUTER_API_KEY with your own API key.

Then completely restart VS Code.

4. Test Claude Code

Open a new VS Code terminal:

claude -p "Reply with exactly: OK"

If you receive:

OK

the connection is working.

Now start Claude Code:

claude

For your first test, use a read-only request:

List the files in the current directory and tell me whether README.md exists. Do not modify anything.

This is a simple way to confirm that Claude Code can access your project before allowing it to make changes.

5. Use Claude Code in VS Code

You can run Claude Code directly from the VS Code integrated terminal and use the Claude Code VS Code extension for a more integrated workflow.

From there, you can gradually experiment with:

Understanding an existing project
Explaining unfamiliar code
Debugging problems
Refactoring code
Writing tests
Implementing small changes
Reviewing your own implementation

Start with small tasks and gradually give the agent more responsibility as you become comfortable with how it works.

6. Keep Codex and Claude Code Separate

If you already use Codex, remember that Claude Code has its own configuration.

Claude Code uses:

ANTHROPIC_AUTH_TOKEN
ANTHROPIC_BASE_URL
ANTHROPIC_MODEL

These are separate from your Codex configuration.

Is It Really Free?

The free usage comes from promotional AI credits, so it should not be considered unlimited or permanently free. Credit availability, model access, eligibility, and usage limits can change.

For learning and getting initial development experience with Claude Code and Opus 5, however, it is a useful opportunity to try an AI coding agent without immediately paying for another subscription.

Start Here

👉 Get Free AI Credits and Try Claude Code

If you are new to Claude Code, this gives you a straightforward way to start experimenting with Opus 5, Claude Code CLI, VS Code, and AI assisted development.

ClaudeCode #Claude #Opus5 #AICoding #CodingAgents #VSCode #AI #SoftwareDevelopment #DeveloperTools

Get $150 in Free AI Credits and Use Codex Directly in VS Code

Md. Shafiuddin Shajib — Wed, 19 Aug 2026 11:33:23 +0000

There is currently a promotional opportunity for developers to claim $150 in free AI credits and use them with agentic coding tools such as Codex.

Claim the free AI credit here: Get $150 Free AI Credit

Get the credit first

Open the link above and create your account.

• Sign up with GitHub when available
• Complete the required account verification
• Open your dashboard and check the promotional credit balance
• Create an API key from the dashboard
• Keep the API key private and never commit it to Git

The promotional amount and eligibility can change, so check the balance shown in your account after registration.

Connect it to Codex in VS Code for windows

Once the account and API key are ready, install the official Codex extension in VS Code.

Open:

VS Code → Extensions → search for Codex → Install

Then configure Codex to use the external provider.

Open by shortcut key windows + R and Search :

%USERPROFILE%\.codex\config.toml

A typical provider configuration looks like this:

model = "gpt-5.6-sol"
model_provider = "agentrouter"
model_reasoning_effort = "medium"

[model_providers.agentrouter]
name = "AgentRouter"
base_url = "https://agentrouter.org/v1"
wire_api = "responses"
experimental_bearer_token = "YOUR_AGENTROUTER_API_KEY"

[windows]
sandbox = "unelevated"

Replace YOUR_AGENTROUTER_API_KEY with your own key.

Save the file, then reload VS Code.

After that, Codex can route your coding requests through the configured provider instead of relying only on a directly billed OpenAI API account.

Why this is useful

For developers working on real projects, this can be useful for:

• Code generation and refactoring
• Debugging existing codebases
• Reviewing architecture and implementation decisions
• Writing tests
• Working across larger project files
• Switching between different AI coding providers when needed

The important part is not simply getting free credits. It is being able to connect those credits to an actual coding workflow inside VS Code.

You can also keep other providers configured and switch between them when your project requires a different model or account.

One practical point

Do not paste your API key into source code, GitHub repositories, screenshots, or public configuration files. Store it securely and rotate it if it is ever exposed.

Start here: Get the $150 Free AI Credit

For developers already using Codex, this is one of the more interesting ways to experiment with free AI coding capacity without immediately adding another recurring subscription.

Full GitHub Guide

For the complete technical setup and reference material, see the companion GitHub repository:

AgentRouter AI Coding Guide

It includes the AgentRouter overview, free AI credit information, GPT 5.6 Sol, Codex, Claude Code, VS Code configuration, OpenAI compatible API details, comparisons, and troubleshooting.

Run DeepSeek Locally in LM Studio: No Internet, No Server Issues – A ChatGPT-Like Experience with a User-Friendly UI

Md. Shafiuddin Shajib — Mon, 03 Feb 2025 17:41:50 +0000

"Run DeepSeek Locally in LM Studio: No Internet Needed, Works on Any PC – A Step-by-Step Guide"

Step 1: Install LM Studio

1️⃣ Visit LM Studio’s official website and download the latest version for your operating system.

2️⃣ Install LM Studio by following the on-screen instructions.

Step 2: Download the DeepSeek Model

1️⃣ Open LM Studio after installation.

2️⃣ Go to the Model Catalog section.

3️⃣ Search for DeepSeek (either the general model or the coding-optimized version).

4️⃣ Choose the appropriate model size based on your system’s capability:

Smaller models (e.g., 7B) for lower-end systems with limited RAM.
Larger models (e.g., 67B) for high-performance setups. 5️⃣ Click Download and wait for it to finish.

💡 Tip: If your PC has lower specs, start with a smaller model for a smoother experience.

Step 3: Load DeepSeek in LM Studio

1️⃣ Once the model is downloaded, go to the Local Models section.

2️⃣ Click on DeepSeek to load it.

3️⃣ Adjust GPU/CPU settings based on your system’s resources.

4️⃣ Click Start Chat to begin using the model.

Step 4: Enjoy an Offline, ChatGPT-Like Experience

✅ No internet or cloud servers required—completely local!

✅ Works seamlessly for text-based and coding-related tasks.

✅ Customize performance settings for optimal speed and efficiency.

🚀 Now, you're ready to use DeepSeek on your PC without any server limitations!