Your coding agent shouldn't stop working because you hit a token limit. You're halfway through a refactor. The agent is making progress. Then the session gets throttled, the meter keeps climbing, or you hesitate to paste client code under NDA into a cloud tool. This 2nd edition book shows you how to run coding agents on hardware you control, with a free companion GitHub repository containing the configuration files, scripts, and examples. Keep the Claude Code interface and workflow. Connect it to an open model running on your own machine. You're running the coding agent with a local model, rather than a cloud model like Claude or GPT. You'll also build local setups with Codex and OpenCode, giving you a choice of workflows for your hardware and coding tasks. Have a local agent editing real files in 15 minutes. Then build the setup you'll want to use every day. Here's what you'll learn: - Code without a cloud token meter. Run suitable tasks against local models and put your own hardware to work.- Keep local inference on your machine. Configure the connection, check where requests go, and test your workflow offline.- Connect Claude Code through a proxy you control. Build the LiteLLM configuration that links the agent to Ollama.- Choose models by what they can actually do. Compare candidates on your hardware and test whether they follow instructions and use tools successfully.- Get useful performance from modest hardware. Tune model size, quantization, and context settings, including a no-GPU laptop configuration.- Fix the failures that make local AI frustrating. Diagnose slow responses, broken tool calls, connection errors, and memory problems.- Build a daily-driver setup. Save your launch commands, prepare a fallback model, and establish working boundaries for agent actions. The companion GitHub repository gives you the framework. Get chapter-organized configurations, startup scripts, the cost calculator, and supporting examples. Run 1 installation command to place the Claude Code proxy configuration and launcher in your local setup, then adapt them to your model and machine. Worked examples show where local models earn their place and where cloud models still offer better results. You'll finish with a repeatable way to choose, configure, and run your own coding stack. Own the setup. Choose the model. Keep building. Scroll up and grab your copy now.
Prijshistorie
* Prijshistorie bevat geen data van Amazon.
Prijzen voor het laatst bijgewerkt op: