I have been using Claude Code with my Obsidian vault for almost a year now. Because it runs locally on my computer, I could potentially interact with these conversations from my mobile using Claude Dispatch, assuming my computer hadn’t gone to sleep.
Yet most of my interaction with Obsidian on my mobile is leaving thoughts, ideas and reminders in a specific inbox note.
That has potentially changed now that Claude Code can run in the cloud, which means I could use it when I’m away from my computer.
This has led me to try to answer two questions:
- What uses, if any, can I make of this new functionality?
- How does Claude Code in the cloud interact with my vault?
The second question is the one I’ll be exploring in this post. The first is the bigger question, and one I might explore in future posts.
Claude in the cloud vs local Claude
When Claude Code runs locally, it runs on my computer’s hardware and has access to specific folders on my computer.
Obsidian, my note-taking and writing application, stores its notes locally as markdown files. A large language model like Claude can read these with ease, so it can access the notes I have created in Obsidian.
When Claude Code runs in the cloud, Anthropic runs up a virtual computer on one of its servers, and Claude Code runs on that. This computer only lasts for the length of the session.
That creates a problem we need to overcome: each conversation is its own sandbox, and by default Claude doesn’t remember things from one conversation to the next.
The obvious answer: GitHub
The obvious answer was GitHub. It’s an industry standard in software development, and Claude Code was initially developed for software developers.
Git is basically a way of storing files that keeps a history of every version, and GitHub hosts these repositories online. Claude could use one as a place to store files, acting as an external memory that lets it remember things between sessions, much as noting something down helps us to remember.
I have set up a GitHub repository, as it’s a place where Claude and I can save files and notes that Claude can use as a form of external memory between sessions.
The answer I already had
Sitting at the back of my mind was how Claude in the cloud could interact with my Obsidian vault. The vault sits at the centre of my work, whether that’s my writing or my learning, so it’s a great source of context for Claude.
It’s also likely that much of what I do with Claude will need to go into my vault, which is becoming a major external memory tool for both me and Claude.
The folder containing my vault syncs with Google Drive, a cloud storage service provided by Google, which Claude can access through MCP.
The first test was to see if Claude could access my existing notes via Google Drive. It could, as it was able to see one of my notes that I had recently updated.
For example, during the initial setup, I mentioned that working on this would make a good blog post, and Claude was able to draft a layout for this post based on that conversation via Google Drive.
Claude wasn’t sure whether the sync would work both ways, so it created a test note in my ZZZZ_AINotes folder and asked me to check whether I could access it in Obsidian. I could, which meant the sync worked both ways. But the vault will only update when my PC is on.
During the test, Claude also noticed that a Git repository had been set up in my vault. I’m not sure how this happened; probably a misunderstanding between me and Claude.
One possible issue is that three tools currently look after that folder: Obsidian Sync, which lets me use my vault in Obsidian on my mobile and tablet; Google Drive; and Git. All of them support versioning. If two devices were working on the same note at the same time, it could lead to duplicate or overwritten notes.
As I’m the only person using Claude and Obsidian in my setup, I consider the risk very small. It’s not the ideal solution, but it’s working for me at the moment.
What does this mean for my red lines?
Claude raised concerns that the fact I could ask it to write something for me on the phone could undermine my red lines. I would like to think that this won’t happen.
Firstly I’m hoping that the practice of writing at my desk directly into Obsidian will be enough to stop this temptation. Secondly, Claude does a good job of considering the possible impact of these red lines.
However, it is a real concern and something I’ll need to monitor going forward.
What’s next
The next step will probably be to explore Git and GitHub. I’d also like to use Claude for other tasks directly related to my writing or knowledge work. I’ll write about them once I’ve explored them.
Have you explored using Claude Code in the cloud? What were your experiences?
You can be kept up to date with our most recent content by signing up to our monthly newsletter.
Further reading
How I use Claude with Obsidian Pt 4
Claude Obsidian Workflow: My PKM Update (Part 3)