This Elasticsearch-based knowledge graph implementation for MCP replaces the previous JSON file-based approach with a scalable, performant solution. It provides distributed storage for entities and relations, advanced search capabilities with fuzzy matching and relevancy ranking, and complete CRUD operations. The system tracks access patterns to prioritize recently viewed and important entities, simulating memory-like behavior where important, recent, and frequently accessed information rises to the top. Built with TypeScript and the Elasticsearch client, it includes tools for migration, management, and maintenance through a comprehensive admin CLI.
No reviews yet. Be the first to review!
Sign in to join the conversation
Highlight AICreate entities with optional observations and reviewInterval
Update existing entities
Delete entities (with optional cascade)
Add observations as separate entities with own freshness
Confirm entity is still accurate, extend review interval
Search with progressive freshness filtering
Get specific entities by name with freshness metadata
Get recently accessed entities
Create relationships between entities
Remove relationships
AI-powered entity retrieval with tentative answers
AI-powered file content inspection
List memory zones (with AI relevance scoring)
Manage memory zones
Manage memory zones
Transfer entities between zones
Transfer entities between zones
Merge zones with conflict resolution
Get entity/relation counts for a zone
Boost entity relevance score
Get current UTC time
