What it does
A memory server for LLM agents built on a local graph. Agents record decisions, tasks, hypotheses and reasoning between sessions, and get a summary of what is open or overdue each time they return to a project. It is a single Rust binary that serves several agents at once, with an optional self-hosted cloud for teams.
Use cases
- 01Keep open decisions and hypotheses between sessions
- 02See which tasks are overdue when a project resumes
- 03Share agent notes with a team on a self-hosted server