I've tried a lot of blogging setups over the years. Dedicated CMS platforms, standalone Markdown editors, even just raw vim sessions committed straight to a repo. They all worked, technically. But all of them had the usual friction, weren't easily automated, and none of them felt like home.
Then last year I started using Obsidian for everything else in my life, notes, projects, daily journaling, research, and a thought hit me: why am I leaving this app to go write blog posts somewhere else?
One Place for Everything
Obsidian is already where I think. It's where I capture ideas at 2am, process meeting notes, and maintain a personal knowledge base that links everything together with internal wikilinks. The Markdown is clean, the editor is fast, and the vault syncs across all my devices.
Hugo, the static site generator that powers my blog, also runs on Markdown. The connection almost makes itself. Other than frontmatter tweaks (e.g. Obsidian frontmatter is yaml while Hugo uses toml), it's nothing at all to just have my blog posting process and pipeline be right there in Obsidian.
So I configured Obsidian as my primary blogging editor. My blog's content lives inside my Obsidian vault as a symlinked folder pointing to the Hugo site's Git repo. The symlink to the repo is not indexed or modified by any Plugins (avoids frontmatter corruption), and I keep a separate "Blog-Posts" folder in Obsidian where I can keep a copy (and backup) of blog content -- while live posts just get saved direct to my blog's Git repo.
So I write, I edit, I organize, all without leaving the app where I already spend most of my working hours. The context switching is reduced, but that lead me to another thought.
What About Agentic AI (Claude, etc) Assistant Roles?
I know every nerd out there seems to be exploring using Agentic AI / LLM systems (OpenClaw, Claude, etc) for not only automation, but for generating content in their voice as well. My initial instinct on this is: "yes to automation, no to content-generation in my voice" as I feel that's lazy, and can lose something. But, truth be told, I already do have my favored AI (Claude Cowork and Claude Code) performing research for me to review. And, yes, sometimes it helps me ideate on drafts, sometiems doing 25-50% of the first-draft. I just can't bring myself to ever do what I'm seeing a lot of YouTube AI influencers seem to be doing: having their AI Agentically transform a braindump into research/first-draft with 100% of the process being attended to by the AI.
So for my own comfort and workflow purposes, once I decided to use Obsidian as my primary blogging interface, I enlisted Claude Cowork & Code (both via their own separate bespoke Skills) to help me automate my workflow once I finish my first draft. This helps me further reduce friction, by letting automate the final step in my blogging pipeline. I do plan later to have it help with each phase of my writing (ideating on raw braindump notes, and introducing rough first drafts, etc). I still want to be the one doing most of the writing (though I have used Claude to do research/iterate on draft revision ideas already in the LLM chat window). There's no reason why Claude can't do a quick check on my final drafts for voice/tone issues or misspellings etc etither. So all that is coming, but for now I've got Claude pushing the content right into my blog's Git repo (Git commit) without any fuss or muss via a scheduled task. And already that's reduced steps for me: I stay in Obsidian, do my writing, indicate when it's ready to publish (via a gate.md style file, that Claude checks if my post is "ready to publish") and boom -- done.
From Obsidian to the Web
Here's how the publishing pipeline works. My blog is a Git repository hosted on GitHub. When I'm ready to publish, the post gets committed and pushed to the repo, and a GitHub Action takes over from there. Hugo builds the site in a few seconds, and the post goes live. No manual FTP uploads, no logging into a web dashboard, no copy-pasting between apps. No database to manage, either, since my blog is just text files. It's literally just me writing Markdown files in Obsidian...then having a process that optimizes them Hugo Markdown frontmatter. In the end, it's just simple and elegant text files driving this blog -- via Claude workflows and Git -- right from my Obsidian app. No "dashboard" or WordPress security nightmares, or Squarespace CMS challenges.
I know this isn't for everyone, but for me the beauty of this setup is its simplicity. Obsidian handles the writing. Git handles the versioning. GitHub Actions handle the deployment. And Claude bridges the Obsidian-to-Git stage in between. Each tool does what it's best at, and I stay focused and spend less time with publishing minutiae.
What's Coming Next
I've been building something on top of this workflow that will add to ideation and content-generation in the future. Again, I'm the writer, it's my voice, but I'm kind of toying with using Claude to help me iterate and refine my ideas into posts. So not just research and editing assistance, but also having jr co-writer around to to handle the tedious parts in my automated blogging pipeline. Claude has demonstrated it can help me stay in Obsidian, stay in-flow, and just do the boring parts of pushing to my site when I'm done. Now I'm going to push further and see how it might help me go from rough idea to published article. But I wholly intend to curate/revise and avoid ever publishing slop! 😂 I want to develop capable assistant that understands my Obsidian vault, my voice, and my process -- but is collaborative early on, while I work entirely on the finished product.
I'll detail that setup in a follow-up post, once I've fiddled and dialed it in. For now, if you're already using Obsidian and you blog with something like Hugo or Quartz (or any Markdown-based SSG), consider bringing those worlds together. The less friction between your thinking and your publishing, the more you'll actually want to draft and refine -- it leaves more time for just writing.
Kind of really the whole point, right?