# Writing

Articles on running AI coding agents in production delivery, drawn from systems Nicholas Hopper has built.

- [Files, not chat: how three agents hand work to each other](https://nicholashopper.com/writing/files-not-chat-how-three-agents-hand-work-to-each-other.md): Nicholas Hopper runs a weekly workflow in which Grok, Claude Code and Codex each own stages and pass work through schema-validated files. This article describes the rules that keep three agents and one person from overwriting each other.
- [What green means when agents ship to main](https://nicholashopper.com/writing/what-green-means-when-agents-ship-to-main.md): When AI coding agents deliver code to a main branch, a passing pipeline is only as good as its definition of passing. Nicholas Hopper has found pipelines that reported success without testing anything, and wrote down what he no longer accepts as green.
- [When the measuring instrument is wrong](https://nicholashopper.com/writing/when-the-measuring-instrument-is-wrong.md): Verifying the work of AI coding agents depends on tests, evals and checkers that can themselves be wrong. Nicholas Hopper has recorded cases where the instrument, not the product, was at fault, and one where an agent changed the instrument to pass.

---

Canonical page: https://nicholashopper.com/writing  
Last updated: 2026-11-17  
Index of all pages: https://nicholashopper.com/llms.txt
