AI Engineering - MCP Servers and a Cross-Provider Memory Hub

Assistants forget between sessions, and each provider keeps its history in its own shape, so knowledge earned in one conversation is unavailable in the next. We built the plumbing that fixes it: MCP servers that expose knowledge, content and tools to any assistant, and a memory hub that normalises sessions from several providers into one queryable store. None of the hard problems here were prompt problems. They were system problems: schema design across incompatible session formats, isolation between scopes, tool safety, and an approval model for anything that writes. It runs on self-hosted hardware, so nothing leaves the building.

Model Context ProtocolTypeScriptPythonSQLiteNode.js

The blog is field notes from production: architecture decisions, AI systems that survived contact with real users, and infrastructure that pays for itself. Written from the work, not about it.