personal· 2026
Moneypenny
Local agent memory and policy controls built on SQLite and sqlite-vector.
Problem
Agents are only as good as the context they carry, yet that context is ephemeral, scattered, and hard to govern.
Approach
A local-first context-management and governance engine for agents: durable, queryable memory on SQLite + sqlite-vector, with a governance layer over what's stored, retrieved, and shared.
What it demonstrates
Databases and data-intensive systems applied directly to agents — landing on the “memory and skills” surface, at the systems layer.