Run Claude Code Locally

You hit the rate limit mid-refactor. Then the monthly bill lands. Again.

Meanwhile, developers on Reddit are running Claude Code on their own machines for about $2 a month. No bill. No rate limits. Nothing leaving the laptop. The difference isn't a better GPU. It's knowing how to point the agent at a model running locally.

This book wires the real Claude Code, Codex, and OpenCode to a model running on your own machine. You'll have a local agent editing real code in 15 minutes. By Chapter 5, Claude Code runs fully local through a proxy you control. By Chapter 9, you'll know your exact break-even month. By the last chapter, your local stack is the daily driver and the bill is dead.

This isn't another "local AI is magic" book. The handful of competitors on Amazon either build a toy agent from scratch or hand you a VS Code plugin. None of them point the actual Claude Code CLI at Ollama. None of them tell you where local loses, or how to fix the 52-second tool call that makes people quit on day one.

Here's what you'll build:

• A local coding agent editing real files in 15 minutes, fully offline.
• Claude Code running 100% local through a LiteLLM proxy you own.
• A model-picker table ranked by what actually worked on your hardware.
• A cost calculator with your personal break-even month, so the $300 bill is gone for good.
• A no-GPU laptop config tuned to run without freezing.
• A troubleshooting runbook for the failure modes that make people quit.
• A locked-in daily-driver stack with a safe local sandbox.