What problem does it solve?
Persistent memory across sessions is hard to maintain without a structured system. PARA Memory Files provides a file-based memory model organized by Tiago Forte's PARA method, spanning three layers: a knowledge graph in $AGENT_HOME/life/ with per-entity folders containing summary.md and items.yaml, a daily notes layer at $AGENT_HOME/memory/YYYY-MM-DD.md, and tacit knowledge captured in $AGENT_HOME/MEMORY.md. It also defines planning rules, memory decay, weekly synthesis, and recall via qmd.
Core Features & Use Cases
- Knowledge graph: store durable facts as atomic YAML entries under projects/areas/resources/archives with summary.md and items.yaml for quick context and on-demand facts.
- Daily notes: maintain raw timeline entries in memory/YYYY-MM-DD.md to capture the day-to-day flow and decisions.
- Tacit knowledge: capture user operating patterns and lessons learned in MEMORY.md for long-term behavior shaping.
- Planning and synthesis: perform weekly syntheses, track fact recency, and ensure no deletions by superseding outdated facts with links to newer ones.
Quick Start
Create a new entity under $AGENT_HOME/life/projects with summary.md and items.yaml, and begin journaling a daily note at $AGENT_HOME/memory/YYYY-MM-DD.md.
What to expect after activation?
This memory model provides a structured, replayable, and recoverable record of knowledge and decisions across time, enabling smarter recall and planning.