---
title: "Comment by Crownforge on \"Stop Choosing an Obsidian Template. Engineer a Vault Architecture.\" | Hashnode"
description: "When an ADR is superseded, how do you flag the runbooks and knowledge notes that still depend on it? Keeping the old decision is useful, but I would want the cu"
url: "https://hashnode.com/posts/article-2026-10-26-0519/6ade649f9a9caf1ce03fbe28/comment/6ae0789987f3dfa3c6294f41"
---

[Discussion](https://hashnode.com/posts/article-2026-10-26-0519/6ade649f9a9caf1ce03fbe28)

Comment [Stop Choosing an Obsidian Template. Engineer a Vault Architecture.](https://hashnode.com/posts/article-2026-10-26-0519/6ade649f9a9caf1ce03fbe28)

[Crownforge](https://hashnode.com/@thecrownforge)

[Oct 27](https://hashnode.com/posts/article-2026-10-26-0519/6ade649f9a9caf1ce03fbe28/comment/6ae0789987f3dfa3c6294f41)

When an ADR is superseded, how do you flag the runbooks and knowledge notes that still depend on it? Keeping the old decision is useful, but I would want the current instructions to show that they need another review

[Daisuke Masuda](https://hashnode.com/@x5gtrn)

[Oct 27](https://hashnode.com/posts/article-2026-10-26-0519/6ade649f9a9caf1ce03fbe28/comment/6ae0789987f3dfa3c6294f41#reply-6ae08370432ab635a1b322cc)

Yes—that is an important distinction. I would keep the superseded ADR as historical evidence, but change its metadata to something like `status: superseded` and `superseded_by: ADR-0042`.

For runbooks and knowledge notes, I use an explicit field such as `depends_on: [ADR-0017]` or a direct link to the ADR. A review query can then surface every current note that still depends on a superseded decision. I would not automatically mark those notes as wrong—the operational guidance may still be valid—but I would flag them as `review_required` until someone verifies them against the replacement ADR.

That preserves the decision trail while making the possible blast radius visible. It is also a good example of why queries should be treated as views: the durable truth is the explicit dependency recorded in each note, while the query simply shows which dependencies now require attention.
