About

Open-source memory and context for user-aware agents.

Remnic helps agents understand the people they work with: their preferences, projects, constraints, decisions, patterns, and definition of good, so they can help with fewer unnecessary questions.

user-aware context layer
Source of truth Scoped memories with provenance
Recall path Relevant context before each turn
Operating model Correctable, bounded, MIT licensed
Why Remnic exists

What Remnic is for

User-aware agents need more than a database of facts. They need a working model of preferences, constraints, decisions, projects, communication style, risk tolerance, definitions of good, and boundaries around where context can be used.

Remnic explores that systems layer: scoped memory, provenance, retrieval quality, correction, boundaries, ask-versus-act decisions, and evals. Local-first storage remains a trust feature, not the whole category.

Maintainer

Who builds it

Joshua Warren is the creator and maintainer of Remnic. Remnic is released under the MIT license, and development happens in public on GitHub.