Show HN: MetaBrain – A local document memory for AI agents

metabrain.eu

8 points by acoye 2 days ago

Hello there HN

I experimented with agentic coding recently and I felt the need to track more contextual data by project. Also I felt the need to be able to go beyond the 1D chat to communicate with agents.

So I created a local document memory, that is discoverable by agents themselves. The CLI is designed to be easy to pick up by agents. It allows humans to collaborate too by reading / searching / editing documents in the store.

I have a Mac native GUI in the review process, I hope it will show up in the App Store soon.

You can try it easily, instructions here: https://metabrain.eu/ Here is the GitHub https://github.com/OpenCow42/metaBrain

The project is also an experiment for me to build some swift project truly cross platform (Mac / Linux / Windows) It is open-sourced with the same license as LevelDB that I wrapped in swift to do this project.

The agents (and humans) can retrieve content quickly with a search, allowing to re-injecting specific knowledge in a specific context during agentic work. It’s funny, I’ve thought of "inference rule base" as something of a derelict idea of the old functional expert systems. Now that I start working with agents I feel more and more the need to go pick previously working solutions dynamically in such a base.

I’d be happy to get feedback. Product fit wise, would this be useful to you or is this just me who is happy with it ?

Finally I had fun with the compression of documents, it tries ZSTD quick, if it does not compress the data by more than 10 percent it stores data uncompressed, else it does a ZSTD level 9 compression on the data. I picked up this trick form OpenZFS.

Thanks

cyanydeez 2 days ago

how is this anything other than advanced context pollution?

  • acoye 2 days ago

    I hear you maybe my summary of the tool does give this impression.

    Here are actual small use cases that work for me:

    - Aggregate the markdown files of a git project in one place. I can version them, review them, dump them, search efficiently into them. Useful to me before launching long agentic tasks, I specify all I want to beforehand.

    - "Cron" checks on new versions of libraries. This works without specific APIs or any setup. A document is maintained with the latest seen version in a specific path of the document base. I also had good results tracking new stars and user adding them with this.

    - Press review / OSINT I built myself a "President Daily Brief" Same thing. It can "reason" given the old daily briefs and assess if the situation is worsening or not on a specific issue, and make basic forecasts. It tracked down the current Ebola crisis before it went mainstream for me and predicted correctly it would worsen.

    I guess I use this for long term memory basically, where I keep ownership of long term knowledge locally. It's more a metaCerebralCortex, and openClaw "dreams" are the hippocampus coalescing these memories at night.

    Also, please note each document in base has a virtual "path", so beyond tags and search you can pin down on a specific domain. It is structured as a tree.

    I do understand that if I were to build a harness this would possibly not be a good fit for the core of it.

    Does this make sense to you, or would you just basically do without such a thing at all to keep each agentic task context clean(er) ?