para-memory-files

Persistent knowledge using the PARA method is organized by storing durable facts in a structured file-based hierarchy across projects, areas, resources, and archives, ensuring long-term context is preserved across sessions.

Daily notes are stored as raw timelines in YYYY-MM-DD.md files, and the system automatically extracts durable facts during heartbeats into a knowledge graph layer for entity-based storage.

File-based memory management works by implementing a layered memory model with an atomic fact store in items.yaml and per-entity folders containing summary.md, supporting recall via qmd queries.

Yes, you can use qmd queries to recall knowledge across memory layers, supporting weekly synthesis, archiving, and retrieval of tacit knowledge patterns stored in MEMORY.md.

You need an environment supporting $AGENT_HOME to implement a PARA knowledge base, where the system creates per-entity folders under $AGENT_HOME/life/ and stores daily notes under $AGENT_HOME/memory/.

Memory decay rules matter because they govern planning and recall workflows by determining how durable facts are retained or archived over time, ensuring the knowledge base remains relevant and uncluttered.