What problem does it solve?
File-based PARA memory provides a structured, durable, and searchable way to store knowledge across sessions. It eliminates scattered notes and manual curation gaps by organizing information into a knowledge graph, daily notes, and tacit patterns, enabling fast recall and coherent planning.
Core Features & Use Cases
- Three-layer memory: knowledge graph with per-entity summary.md and items.yaml, daily notes as a raw timeline, and tacit knowledge about user patterns.
- Lifecycle rules: save durable facts to items.yaml, weekly synthesis to refresh summary.md, and supersede outdated data without deletion.
- Planning and recall: plan with timestamped notes in memory/plans and recall past context via qmd or semantic search.
Quick Start
Organize memory with PARA by initializing the life/ areas/ resources structure under $AGENT_HOME and start saving durable facts to items.yaml for long-term recall, then perform weekly synthesis to refresh summaries.