The biggest public thing I'm working on is an AI sandbox tool that keeps it separated from your system and secrets.

Changes the AI makes to the "workdir" are extracted via diffs or git commits, so you can see what the AI did before deciding if you want it or not.

It has no access to your home dir, no access to your env, and you can restrict its network access.

Containment can be container or VM level, with a Linux or MacOS contained environment.

https://github.com/kstenerud/yoloai

Tinkering and thinking about the hacker news data in relation to some machine learning ideas.

Here's my data and search engine with all the HN data:

* https://orangewords.com/

A project where you create websites by just using the chat interface of your llm. It's like a paas but oriented to llms. If your ui support mcp It should work.

There are some examples I did with opus, sonet and deepseek models: https://onvibe.run/forkable

It has pg, file uploads, emails, etc. The mcp provides tools to run migrations, snapshots, drafts, etc.

I started It to allow my wifi and daughter create their custom tools.