Skip to content

How-to guides

Task-oriented. Recipes for solving specific problems the reader brought with them. Assumes baseline competence; doesn't teach.

Writing a how-to

A good how-to:

  • Solves one named problem. The title is the problem: "Configure X for production", "Migrate from Y to Z", "Debug a failing W".
  • Assumes the reader is competent. Doesn't reteach the basics. Links to tutorials for foundational concepts.
  • Is goal-oriented. Skip context that isn't needed to accomplish the goal. The reader is here to do something, not to learn.
  • Handles the realistic version. Cover the common variations and pitfalls — that's what makes how-tos different from "just read the reference."
  • Is named for what the reader was searching for. "How to configure rate limiting" beats "Rate limiting configuration guide."

What goes in a how-to

  • A clear problem statement at the top.
  • Prerequisites and assumptions.
  • The steps to accomplish the goal — terser than a tutorial.
  • Variations the reader is likely to need.
  • Pitfalls and how to recognize them.
  • Links to relevant reference material.

What does NOT go in a how-to

  • Step-by-step beginner instruction. That's a tutorial.
  • Complete authoritative description of every option. That's reference.
  • Why-this-design explanations. That's an explanation page.

Maintenance

How-tos drift when the product changes underneath them. Make doc updates part of the spec workflow: when a spec ships, check whether any how-to references the changed behavior, and update in the same PR.

Pages in this directory