آموزش نصب و استفاده رایگان از Claude Code

Summary

Mohammad Dehdar presents a tutorial on using Claude Code without a Claude account by connecting its interface to other AI models through a GitHub-hosted proxy.

Setup

1. Install the proxy. Follow the instructions in the linked GitHub repository’s README. The presenter uses a PowerShell command on Windows; Linux users are directed to a separate command. The installation handles the required dependencies.
2. Start the local dashboard. Run the proxy’s server command to open a localhost admin dashboard. Use it to configure providers and models without editing project files directly.
3. Connect OpenRouter. Create an OpenRouter API key, enter it in the dashboard, validate it, and refresh the model list.
4. Choose and configure models. Select models from OpenRouter’s free-model listings and assign them to Claude Code’s model settings. Use the correct OpenRouter provider name for validation. The presenter recommends an NVIDIA model for coding, while noting that coding-optimized models are preferable.
5. Use Claude Code in VS Code. With the server running and the Claude Code extension installed, restart VS Code. The presenter demonstrates asking the assistant to inspect a project and make changes, then approving its requests to access or edit files.

What “free” means

Claude Code provides the coding workflow and interface, but responses come from the selected third-party models—not Claude’s own model. Free-provider limits may apply. The presenter suggests creating another OpenRouter account after reaching a limit and promotes Atomic Mail as a way to create email addresses for additional accounts.

Other providers and resources

The repository reportedly supports several providers, including OpenRouter, NVIDIA, Google AI Studio, and DeepSeek, with a similar setup process.

Main speaker: Mohammad Dehdar Supporting resources: The GitHub repository README and OpenRouter