Simon WillisonFeb 23
I've published the first two chapters of a new guide to Agentic Engineering Patterns - coding practices and patterns to help get the best results out of coding agents like Claude Code and OpenAI Codex https://simonwillison.net/2026/Feb/23/agentic-engineering-patterns/
Writing about Agentic Engineering Patterns

I’ve started a new project to collect and document Agentic Engineering Patterns—coding practices and patterns to help get the best results out of this new era of coding agent development …

Simon Willison’s Weblog
Show threadSimon WillisonFeb 23
"Writing code is cheap now" talks about the central challenge of agentic engineering: the cost to churn out working code has dropped to almost nothing, how does that impact our existing intuitions about how we work, both individually and as a team? https://simonwillison.net/guides/agentic-engineering-patterns/code-is-cheap/
Writing code is cheap now - Agentic Engineering Patterns

Writing code is cheap now - Agentic Engineering Patterns

Simon Willison’s Weblog
Show threadSimon Willison
"Red/green TDD" talks about how you can get much better results from most coding agents by encouraging them to use test-first development where they watch the tests fail before building an implementation that lets them pass https://simonwillison.net/guides/agentic-engineering-patterns/red-green-tdd/
Red/green TDD - Agentic Engineering Patterns

Red/green TDD - Agentic Engineering Patterns

Simon Willison’s Weblog
Feb 23 at 5:54pmWeb
Show threadStephan SvobodaFeb 23
@simon i also added to have 100% coverage for new Code
Without that my gemini agent used to add unnecessary stuff
Show threadvmaurinFeb 23
@simon So we have to relearn everything. Soon "Breaking news: having too much lines of code is bad, how to train your agent". God ...
Show threadjtonlineFeb 25
@simon I've been arguing that it's more important to understand and verify the tests actually test what they claim to be testing than to understand the AI generated implementation. It's fair to say that this appears to be a controversial view!
Show threadGlenjaminMar 4

@simon which model/harness have you been doing this on?

Whenever I try it, I just get one big dump of test code, then one big dump of implementation code afterwards - and the Claude plan.md usually has a sketch of the implementation already by default