Architecture proposal: how Synto fits into a unified local knowledge OS

[noirblue]

Hi kytmanov,

First: Synto is excellent. I used it extensively while trying to build a unified Jarvis memory stack, and it was the best-in-class solution for [specific thing: compilation / audit / conversation / memory lifecycle].

After weeks of integration work, I wrote a specification for what a unified local knowledge OS would look like — not to replace Synto, but to give it the system layer it deserves. The architecture explicitly preserves each tool's core innovation:

• Synto's two-tier compilation and hand-edit protection
• Synthadoc's multi-format ingest and claim-level audit
• LLM-WIKI-MCP's conversation history and context budgets
• Link's graph context and memory lifecycle

The spec is here: https://github.com/IsaacCLupus/mnemosyne-spec

And the comparison with existing tools: https://github.com/IsaacCLupus/mnemosyne-spec/blob/main/COMPARISON.md

I'm not a developer; I'm a systems thinker who validated the integration pain the hard way. I'd love your feedback on whether the architecture properly captures what Synto does best — and whether there's any interest in collaboration or an import adapter down the road.

No pressure, just wanted to share the blueprint with the people who built the foundations.

Regards,
noirblue

p.s. This project is named in memory of my brother, Isaac.