Access the graph from multiple machines

[futureformed]

I have my main coding machine, and I intend on installing Graphify. However, when I'm on the move, I will sometimes code or want to access information via my laptop. For someone who works like this, what's the best way forward? Is there a way of building the graph using the main machine and then hosting it somewhere so the laptop can access it also and contribute towards it?

[daveshenal]

Hi futureformed,

This is actually exactly how Graphify is designed to be used, no separate hosting/sync setup needed.
The graph isn't stored anywhere external, it's just files under graphify-out/ in your project folder (graph.json, GRAPH_REPORT.md, the HTML visualization, etc.). The recommended pattern is to commit graphify-out/ to git alongside your code:

1. On your main machine, run /graphify . (or graphify from the CLI) to build the initial graph, then commit graphify-out/ along with your code.

2. On your laptop, just git pull (or clone) as normal - you get the graph for free, no rebuild needed. Your AI assistant reads GRAPH_REPORT.md immediately.

3. To keep it current as you code on either machine: run graphify hook install once, which sets up a git post-commit hook that rebuilds the graph automatically after code changes - no LLM calls needed for code-only updates, so it's fast and free.

4. For doc/paper/image changes specifically (which do need an LLM call for semantic extraction), whoever edits those files runs /graphify --update before committing.

A couple of things worth knowing if both machines might run extraction at different times:

graphify-out/cache/ and graphify-out/manifest.json are meant to be gitignored - the cache is mtime-based and invalid after a fresh clone anyway, and the manifest is just local token-usage tracking.

If both machines happen to commit graph changes around the same time, you can get graph.json merge conflict markers. Run graphify hook install (mentioned above) - it also sets up a git merge driver that union-merges graph.json automatically, so this shouldn't bite you in normal use.

So no need to "host" anything separately... git itself is the sync mechanism, and the hook handles keeping things current on whichever machine you're on.

[futureformed]

This is awesome - thanks @daveshenal

I'm mainly doing knowledge work, regularly adding new documents and transcripts, etc., to my Google Drive and local drive. Would it be possible to run a regular job to automatically push that stuff to the Github? I guess it could, that way. Every day at the end of the day, I could synchronise my most recent documents to Git and then download them to my local machine when required.

I'm just wondering how this works if I'm using the iOS or iPadOS versions of Claude, for example. Maybe this is a bit of an edge case, as I don't do a huge amount of knowledge work on mobile devices. Maybe the GitHub connector?

[daveshenal]

Glad it helped! For the sync job... yes, this is doable, but worth knowing how it actually runs: Claude Cowork has scheduled tasks (/schedule in a Cowork session) that can use a Google Drive connector + GitHub connector to push new docs to a repo daily.
One catch: the schedule runs on your desktop app, not your phone - your Mac/PC needs Claude Desktop open (just logged in is enough) at the scheduled time, or it skips and catches up next time you open it. So iOS/iPadOS works for triggering a run or checking results, but the actual daily job needs a desktop machine staying on.
Bit of a tangent from Graphify itself, but figured I'd answer since you asked... for Claude.ai-specific stuff like this going forward, Anthropic's support docs are the more reliable source than this repo.