I've been using Claude Code for non-coding tasks, like sorting emails and analyzing my spending This was annoying to setup because Claude doesn’t have the right context - it can’t see my email or my credit card statements So we built a virtual filesystem that turns all of my... data sources (gmail, notion, gdrive and more) into local folders on my computer I can just describe what I need (e.g. "invoices from the last week") and a folder appears on my computer with exactly those files This gives Claude context, but it also gives it memory - since it has a local filesystem that syncs in the background, it continuously gets access to my data The more things you expose as files, the more Claude can do for youshow more

I did something very similar today as I wanted my agents to have their own email, I added google workspace and then added @claudeai and approved it. Now it can do that, or view my files directly as well as managing the company email accounts. [email protected] lands on my Ivy agent and she connected to that box and the support box, also my personal box where she sorts it all before briefing me on it and my calendar for the day. lol I love this so MUCH~

@WhiteFlourFL @claudeai ❤️

I see there is a self host option? So it’s all open source and we can run locally?

Yup it’s 100% open source

Open-source, locally mounted files keep Claude practical while preserving user control

You can give Claude code access to these things using skills and API’s Just so you know

Yes you can use APIs to get data it’s true

I should have watched your video before saying anything. Very cool dude. Thanks for sharing

💪

@imjaredz nice, really interesting idea

@imjaredz 🙏🙏

@imjaredz no API?

@imjaredz we have an API, DM me

The amount of times I’ve fought myself on building this! The UX seems incredible, anyone who doesn’t see the power in this is missing out big time. Great stuff mate 🙏

really smart, do you think you'll need/want a sync from tool>filesystem (with LLM filtering perhaps) in future. Or does this one time grab fit most use cases?

we automatically refresh the data each each time you access a file each folder you create is filtered by an LLM, based on the “smart query” you create

Continuous data refresh with a unified namespace makes Claude practical at scale

Drive seems like the best context

this is amazing

Thanks Kevin! 🙏🙏

Very cool, dude!

the continuous sync is what makes this actually useful. claude getting fresh data each session without manual setup is huge

Smart

Expose is the key word here.

This is fundamentally so cool

Virtualizing data as a local filesystem is the ultimate context injection. You didn't just give Claude memory; you gave it an OS.

You can find many more similar use cases here:

This feels like the missing abstraction. Files are the universal interface agents already understand.

This 👇

This is how clawdbot started

This is brilliant, and amazing that it can all be run locally.

I tried something similar with local project docs, and it made a bigger difference than I expected. Before building this, what was the biggest pain you personally kept running into?

Since you're using Claude Code, make sure to add a file to the root of your virtual filesystem. In it, define your "Life Style Guide." Tell it how you prefer your spending summarized (e.g., "Always prioritize tax-deductible categories") or how you want your Notion pages formatted. This ensures that every time you run a command, Claude already has your "Executive Intent" in mind.

The human is exposed as the weak link in this transformation. The first recommendation should be a context for the originator to guide all the downstream plans and actions. Claude is ready, willing and able to formulate generative productivity. It requires the human to learn how to shed specialist thinking and grasp the potential of collaborative engagement.

The install script is not resolving currently, too popular?

that URL is wrong! should be

>This gives Claude context, but it also gives it memory have you seen it actually use historical context from previous sessions effectively? Like does it remember patterns in your spending or email habits across different convos?

@threadreaderapp unroll

I guess you don't know, but Gmail has an api (and a batch api you definitely should use) I used it to archive something like 15 years of emails from Gmail into local, highly compressed archive files, split by month and stripped of all useless newsletters and such. They are now encrypted and sitting on multiple cheap cloud storage servers. Cloudflare R2 and Backblaze are good options.

Reminds me of FUSE in Linux

I use Gmail mcp for that

nice. Interested to see if it can seamlessly be used with Obsidian (and opencode)

So the files are created and write text inside really?