It's interesting that I've also gone down a similar journey and the same keywords, fences, gates .. also showed up in my ai government platform.

yea - that's exactly what i saw. I like claude code, but err.... I wish they allowed better auditability. I'm using headroom, so i see somethings but I can't relay it back to what anthropic has.

2 factor vote. Vote in app, still go in person to validate result.

Here's me playing around with it. https://www.youtube.com/watch?v=2oJohBiqMUA

My approach/prototype using XState with websockets from an MCP server https://github.com/uptownhr/mcp-agentic-ui

appreciate it.

On why I built it and how. https://uptownhr.com/blog/building-lattice-knowledge-graph-c...

I also built something similar. Utilizes ZEP for building a temporal graph that keeps track of facts.

https://github.com/Zabaca/temporal-bridge

Did some research and guru is in the realm of solving this with ai

That is true, but feels like a miss as it's a different specialty to write good docs vs organizing and grooming docs.