Claude Code: Up and Running

Chapter 11. Beyond Claude Code

A Note for Early Release Readers

With Early Release ebooks, you get books in their earliest form—the author’s raw and unedited content as they write—so you can take advantage of these technologies long before the official release of these titles.

This will be the 11th chapter of the final book. The GitHub repo is active.

If you’d like to be actively involved in reviewing and commenting on this draft, please reach out to the editor at [e-mail].

Chapter 1 drew a line between the model and the harness. This chapter is where the two finally come apart. We start with the surfaces, the places other than a terminal where the same harness shows up, and with what happens when you open several of them at once. Then we look at cost and at which model belongs on which task, and I’ll tell you what I actually do.

Then comes the part that surprises people. Claude Code can run against models that do not come from Anthropic. Anthropic documents the knobs and says plainly that it does not support routing Claude Code to non-Claude models, but once you have an API key and set a URL, the harness will talk to something else. We’ll use OpenRouter as a near-universal API for hundreds of models and watch Ollama do the same job under the hood in one command.

After that we run the harness in a loop. You give a high-level goal and an evaluation function, and we climb that ladder a rung at a time, until the loop moves off your machine and becomes an always-on ...