From prompt to production

Three steps from a prompt that happened to work to one you can hand to a team: give it a reusable structure, write it down where the agent will read it, and prove it still works after you change it.

1. Part 1

   How to write a prompt you can reuse

   Most prompts are lucky one-offs. The ones that keep working share a structure: an explicit role, a checklist of constraints, a worked example, and a self-check step. Here's how to build one.

   Jun 18, 2026
2. Part 2

   The instruction file is the highest-leverage file in your repo

   AGENTS.md, CLAUDE.md, copilot-instructions.md and .cursor/rules all solve the same problem. Most of them are written badly. Here's what changes agent behaviour and what's decoration.

   Jun 26, 2026
3. Part 3

   Ship a prompt like code: a five-case eval harness you can build in an hour

   You wouldn't ship a function you ran once. Here's the smallest evaluation setup that catches real regressions — five cases, three graders, and a rule for when to add a sixth.

   Jun 29, 2026