The Obsidian Vault Structure I Use to Run 5 Blogs Solo

In my last post I said the “second brain” was half of how I run five blogs alone. That post was the map. This one opens up the part readers asked about most: what the vault actually looks like inside.

I run all five blogs out of a single Obsidian vault. Not five vaults, not five apps. One. The structure is what stops that from turning into a junk drawer.

Quick conclusion

One vault can hold five blogs if you give it three layers:

Project separation. Every blog gets its own folder with the exact same inside shape, so switching blogs feels like switching tabs, not switching tools.
Shared resources. Anything used by more than one blog (my profile, reusable procedures, templates) lives in common folders I write once and reuse everywhere.
A per-article workspace. Each article is one folder that moves through the same five stages, from research to post-publish check.

On top of those three layers sit a few plain text files that let me, or an AI, reopen any blog cold and keep working. That last part is the whole point.

What this article covers

Why I keep one vault instead of five
The three layers of the structure, folder by folder
The files that make a blog resumable from a cold start
What broke along the way
The smallest version you can copy today

I’ll save the exact configs and prompts for a later, more detailed write-up. This is the structure, not every setting.

Why one vault, not five

When I started, each blog lived in its own space. It felt tidy. It wasn’t.

Three problems showed up fast:

Context got stranded. A note I wrote for one blog was invisible to the others, even when the lesson applied to all of them.
Rules drifted. Each blog had its own copy of “my” style and SEO rules, and they slowly fell out of sync. Fixing one meant remembering to fix four more.
Templates multiplied. I kept rebuilding the same article folder by hand, slightly differently each time.

One vault fixes all three. Shared lessons live in one place. Rules have a single source. A template is written once. The cost is that you have to be disciplined about structure, because there’s no longer a wall between blogs to keep things sorted. The three layers below are that discipline.

Layer 1: Project separation

At the top level, each blog is one project folder. I name them so they sort together, like Project_<BlogName>. The key rule: every project folder has the same inside shape.

Inside each one:

AI context. The instruction file I hand to an AI before any work: voice, rules, hard “never do this” lines.
Knowledge. The article rules, the style guide, the SEO notes for that blog.
Session logs. What I did and what I learned, dated.
Workspace. Where articles actually get built.
Briefings. The weekly plan.
Settings. Fixed pages and site config.

Because the shape never changes, my brain doesn’t have to reload. I open any blog and the instruction file is always in the same spot, the rules are always one folder over. Switching from one blog to another costs almost nothing.

Layer 2: Shared resources

This is the layer that makes five blogs feel like one workflow instead of five.

Anything more than one blog needs lives in shared folders that sit beside the projects, not inside them:

Common context. A master profile and shared character notes. Who I am, who I write as, the things every blog should know about me.
Reusable skills. Saved procedures the AI follows: set up a workspace, write a draft, fact-check, prepare the export, sync after publishing. I write a procedure once and every blog uses the same one.
Templates. The starting shape for new notes.
System. Cross-blog governance and logs.

The rule I follow: if I’d copy-paste it into a second blog, it doesn’t belong inside a blog. It goes in a shared folder. That single rule is what keeps the vault from sprawling.

Layer 3: The per-article workspace

Down inside each blog’s workspace, every article is one folder, and that folder always moves through the same five stages:

1_research: notes and sources
2_drafts: the first draft and the post settings
3_reviews: the fact-checked, cleaned-up version
4_exports: the final file I paste into WordPress
5_check: the post-publish review

The numbers matter more than they look. They force the folder to read top to bottom in the order the work happens, so I can never confuse a rough draft with the final file. The export folder holds the one true version. Everything before it is just history I can throw away later.

The files that make a blog resumable

The folders are the skeleton. Three kinds of files are what let me reopen a blog after two weeks, or hand it to an AI, without re-explaining anything.

An entry point. A short guide at the top of the vault, plus an instruction file per blog. This is what an AI reads first: which blog am I working on, and what are its rules?
Knowledge files. The rules and style guide live here, in text, not in my head. That’s why the voice of each blog stays steady even when I’m away from it for a while.
A handover note. Every article folder has a one-page HANDOVER.md. At the end of each stage, whoever did the work updates it: what’s done, what’s next, who’s next.

That handover note is the small file that does the heavy lifting. Because of it, a cold start isn’t really cold. The vault remembers, so I don’t have to.

What broke along the way

I’d rather show the dents than pretend the structure arrived clean.

A category got out of sync. A draft’s metadata listed one category while the publish settings listed another, and the mismatch slipped through to a live page. The fix was a rule: lock the category before the export stage, not after.
Backup folders piled up. Saving a snapshot before each stage is safe, but those snapshots stacked up until a workspace was hard to scan. Strict date-and-slug naming is what keeps them findable instead of scary.
Names drifted across files. One article’s headline didn’t match the text baked into its cover image. Now I decide the title in one place and copy it everywhere, instead of the other way around.

None of these came from the structure being wrong. They came from me not following it closely enough. The structure was the fix, not the cause.

What you can copy today

You don’t need five blogs or any custom tooling to get most of this. Start small:

Pick one app as your single source of truth. Obsidian, or whatever you’ll actually open every day.
Give each blog the same folder shape. Even just AI context, knowledge, and a workspace.
Write one instruction file per blog. Voice, rules, things to never do. Hand it to your AI every time.
Make every article one folder with stages. Research, draft, review, final. The final folder holds the one true version.
Keep a one-page handover note. Update it whenever you stop, so the next session starts warm.

That’s the 20% of the structure that delivers most of the result.

FAQ

Doesn’t one vault for five blogs get messy?
It would, without Layer 1. The same folder shape in every project and a strict rule about what goes in shared folders are what keep it sorted. The vault is big, but it’s never confusing.

Why number the folders, like 1_research?
So the folder reads in the order the work happens. Numbering also stops me from ever mistaking a rough draft for the final export, which is the one mistake that actually reaches readers.

Do I need plugins to make this work?
No. The structure is just folders and plain text files. The only thing I rely on is keeping that folder synced across my devices through ordinary cloud storage. Plugins are a nice-to-have, not the foundation.

What if I only have one blog?
The same shape still pays off. The per-article folder and the handover note alone will save you every time you come back to a half-finished post.

Next steps

That’s the skeleton: one vault, three layers, and a handful of files that make any blog resumable. It’s the part of the “second brain” I lean on every single day.

The next deep dive goes to the other half of the system: how I split the actual work across different AI tools, and which one I trust with which job.