FREE Claude Code Setup: Step-By-Step In 10 Minutes

FREE Claude Code setup is straightforward if you do the steps in the right order.

This post walks every command, every config file, every "wait what?" gotcha.

By the end you'll have Claude Code running through a free model proxy.

10 minutes start to finish.

What You'll Have After

Free Claude Code repo cloned
.env configured with Nvidia NIM key
Proxy server running on localhost:8082
Claude Code launched with custom env vars
A first successful AI coding test

Total cost: £0.

🔥 Want my full FREE Claude Code setup video walkthrough + troubleshooting tree? Inside the AI Profit Boardroom I've recorded every step on screen, documented every error I've hit, and the diagnostic tree for what to do when things break. 2,800+ members already running this. Click below. → Get the setup walkthrough

Prerequisites

macOS, Linux, or Windows with WSL2
Python 3.11+
uv installed (fast Python package manager)
git
A working terminal

If uv isn't installed:

curl -LsSf https://astral.sh/uv/install.sh | sh

Restart terminal. Verify with uv --version.

Step 1 — Clone The Repo

Open terminal. Pick a directory.

git clone https://github.com/[free-claude-code-repo].git
cd free-claude-code

Repo is on your machine.

Step 2 — Get An Nvidia NIM Key

Open browser. Go to Nvidia NIM signup.

You get a free API key with 40 requests/minute.

Copy the key.

Step 3 — Configure .env

cp example.env .env
nano .env

(Or open .env in any editor.)

Paste your Nvidia NIM key:

NVIDIA_API_KEY=nvapi-your-key-here

Set model mappings:

MODEL_OPUS=nvidia_nim/Z-AI/GLM-4.7
MODEL_SONNET=nvidia_nim/Z-AI/GLM-4.7
MODEL_HAIKU=nvidia_nim/Z-AI/GLM-4.7

Save. Close.

Step 4 — Run The Proxy Server

uv run uvicorn server:app --port 8082

Output:

INFO:     Started server process [...]
INFO:     Uvicorn running on http://0.0.0.0:8082

Server is listening.

Leave this terminal tab open.

For more on the proxy architecture, my build your own openclaw post covers similar harness patterns.

Step 5 — Launch Claude Code With Env Vars

Open a new terminal tab.

ANTHROPIC_BASE_URL=http://localhost:8082 ANTHROPIC_AUTH_TOKEN=ccc claude

Claude Code launches.

Test:

> Build a simple Python script that prints hello world

Claude Code responds, generates the script, saves it.

If you see a response, setup is verified.

Step 6 — Make It Permanent

Don't want to type those env vars every time?

Add to ~/.zshrc (or ~/.bashrc):

alias freeclaude='ANTHROPIC_BASE_URL=http://localhost:8082 ANTHROPIC_AUTH_TOKEN=ccc claude'

Save. Restart terminal.

Now freeclaude launches Claude Code through the free proxy.

Step 7 — Add VS Code Integration (Optional)

Open VS Code settings.json.

Add:

{
  "anthropic.baseUrl": "http://localhost:8082",
  "anthropic.authToken": "ccc"
}

VS Code's Claude integration now uses your free proxy.

For more on VS Code AI patterns, my deepseek openclaw post covers the IDE harness.

Step 8 — Discord Bot (Optional, 10 minutes)

Inside the repo, follow the Discord bot setup:

Create Discord application
Invite bot to your server
Set bot token in .env
Run bot startup script

Now you can ping the bot in Discord with prompts.

Voice notes work — auto-transcribed and routed.

For Telegram bot patterns specifically, my hermes deepseek post covers similar setup.

Common Setup Issues

Port 8082 already in use. Change port in run command + Claude Code env var.

Nvidia API rejected. Re-copy key — sometimes invisible chars get pasted.

uv: command not found. Install uv (see Prerequisites). Restart terminal.

Claude Code doesn't connect. Make sure proxy server is running in another tab.

Rate limit hit immediately. You exceeded 40 req/min. Wait 1 minute. Add per-request batching.

For broader diagnostic patterns, my hermes ai course post covers troubleshooting trees.

🔥 Want my .env template + Discord bot config + production setup? Inside the AI Profit Boardroom I've put up the bash automation script, the .env templates for each provider, and the bot configs. 2,800+ members already running these. Click below. → Get the production setup

Switching To OpenRouter

If Nvidia rate limits hit too often:

MODEL_OPUS=openrouter/glm-4.7-free

Get an OpenRouter API key (free tier available).

Add to .env:

OPENROUTER_API_KEY=or-your-key

Restart proxy server. Claude Code now uses OpenRouter free tier.

Switching To LM Studio (Local)

For full offline / fully free:

Install LM Studio
Download GLM-4.7 weights
Start LM Studio's local server
Configure proxy:

MODEL_OPUS=lmstudio/glm-4.7

Now everything runs on your machine. No external API calls.

Best for privacy. Requires decent GPU.

Maintenance

Update repo weekly. git pull to get latest. Active project.

Rotate API keys quarterly. Hygiene.

Monitor rate limits. If hitting limits often, switch provider or upgrade.

For maintenance discipline, my hermes ai course post covers the patterns.

FREE Claude Code Setup FAQ

Do I need a powerful machine?

Proxy is light. LM Studio path needs GPU. Nvidia/OpenRouter paths are cloud-based.

Can I run on a server?

Yes — VPS works. Configure HTTPS via Caddy or Cloudflare.

Multiple users?

Yes — multi-user deployment possible. Each user with their own Claude Code launch alias.

Costs?

£0 if using Nvidia/OpenRouter free. £5-10/month VPS if hosting remote.

Onboarding for team?

15-30 minutes per person. Most of that is signup for Nvidia.

Can I bill clients for time saved?

Yes — many agencies bill AI-coding time as productivity.

Final Take

FREE Claude Code setup is 10 minutes for a free Claude Code stack.

Six commands.

One config file.

First AI coding session live.

Stop reading and install.

🔥 Ready to ship FREE Claude Code tonight? Get a FREE AI Course + Community + 1,000 AI Agents 👉 join here. Or grab the production setup inside the AI Profit Boardroom.

Learn how I make these videos 👉 aiprofitboardroom.com

Video notes + links to the tools 👉 skool.com/ai-profit-lab-7462

Free claude code setup is 10 minutes — go and ship it.