The API docs pipeline that runs itself (week 1 recap)

#api #tooling #documentation #devops

One week in. Let's connect the dots, because the tools are useful alone but the real win is chaining them.

Here's the pipeline:

You build and test an endpoint. You verify it with cURL. You run that through cURL to Markdown (https://markdown.co.in/tools/curl-to-markdown.html), or generate from your OpenAPI spec (https://markdown.co.in/tools/openapi-to-markdown.html) if you're spec-driven. You add the human layer: a plain description and a working first example. Then you keep it honest by regenerating in the same PR.

The whole point is that accuracy stops depending on memory. Your docs come from the things you actually ran. That's the difference between docs you trust and docs you tolerate.

Mindset shift for the week: documentation isn't a chore you do after the real work. It's a byproduct of doing the real work well.

Next week: conversion and migration. Moving between HTML, Markdown, and PDF without losing your weekend.