Claude Code: Up and Running

Chapter 10. Review, Testing, Debugging, and Failure Recovery

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 10th 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].

Claude Code can produce a great deal of code quickly, and that changes the economics of software work. It does not repeal the oldest rule of software engineering, though. Code is not done because it exists. It is done when it behaves correctly and someone can still maintain it. For a risky change, it is done when the evidence says so.

At skyscraper scale, generating code is the easy part. Controlling long-term complexity is the real work. An agent that implements a feature today and quietly weakens the tests has not saved anyone time. Neither has one that leaves the docs describing yesterday’s behavior. The work has been deferred to whoever touches the system next, and increasingly that is another agent session with none of this one’s context. This chapter is about habits that scale with the code.

Code Review and Evidence Governance

Most of this chapter is about review, because in my experience review is where the speed of agentic coding either pays off or gets ...