map-connections

To build a memory graph from markdown files, run a scan against the people/, projects/, and context/ directories. The tool extracts entities, infers relationships, and stores the structured graph idempotently in memory.

Yes, you can update your memory map using incremental scanning. The tool detects changes since the last run and processes only modified markdown files, keeping the knowledge graph up-to-date without rescanning everything.

Entity extraction and relationship inference parse markdown context files to identify people, projects, and organizations, then deduplicate the results and map connections to build a structured knowledge graph.

Yes, the memory graph applies deduplication to people and projects found across multiple markdown files. Identical entities are merged and stored idempotently to prevent duplicate nodes in the relationship map.

The best way to track relationships in markdown is to aggregate context files into a memory graph. This approach automatically infers connections between people and projects, providing a live map for context recall.

You should avoid incremental scans and perform a full scan when structural changes occur in the markdown files or when the memory graph needs to be rebuilt from scratch to ensure all entities and relationships are captured.