Authoring Kit

How MDS uses AI agents to keep its documentation clear, accurate, and up to date.

The content quality dashboard tells you how well MDS documentation is scoring today.

The authoring kit is what gets it there: the rubric, agents, and workflow the MDS team uses to write new pages and improve existing ones, working alongside the people who own the content.

Why this exists

Good documentation is hard to keep good. Pages drift out of date, wording that made sense at the time turns out to be ambiguous, and a busy team can’t manually re-read hundreds of pages every time the product changes. The authoring kit turns “is this page still clear and correct?” into something that can be checked consistently, at any time, rather than relying on someone remembering to look.

The MDS team stays in charge

The authoring kit takes on the repetitive parts of keeping documentation good: researching prior art, drafting, and scoring, freeing the MDS team to spend its time on judgment calls. Every draft the kit produces is reviewed by a person before it goes anywhere near a live page, and publishing a page is always a human decision.

What’s inside

A rubric

A shared standard for what “good” documentation looks like: clear, unambiguous,and useful to a person and to an AI assistant reading the same page. Every score on the content quality dashboard comes from this rubric, so every page is held to the same bar.

A team of AI agents.

The kit splits the work across specialists that support the MDS team: one that researches how other design systems handle a topic, one that writes the page, one that scores it independently against the rubric, and one that coordinates the other three. Splitting the work this way keeps each agent focused and keeps the scoring honest, because the agent that writes a page is never the one that grades it.

How a page gets better

When someone on the MDS team asks the kit to improve a page, the agents run a loop: research the topic, write or revise a draft, score it against the rubric, and repeat until the page clears the bar or the agents run out of ideas worth trying. A person always reviews the result before anything goes live, and publishing the change is always the MDS team’s call.

Browse the kit

  • Agents: meet the specialists that work together on documentation.
  • Workflows: follow how the agents research, write, and review.
  • Skills: explore the guidance and references they use.
  • Review: read the rubrics behind content-quality assessments.