JSPM

@grackle-ai/knowledge

0.91.2
  • ESM via JSPM
  • ES Module Entrypoint
  • Export Map
  • Keywords
  • License
  • Repository URL
  • TypeScript Types
  • README
  • Created
  • Published
  • Downloads 271
  • Score
    100M100P100Q137436F
  • License MIT

Knowledge graph subsystem for Grackle (Neo4j + vector search)

Package Exports

  • @grackle-ai/knowledge
  • @grackle-ai/knowledge/dist/index.js

This package does not declare an exports field, so the exports above have been automatically detected and optimized by JSPM instead. If any package subpath is missing, it is recommended to post an issue to the original package (@grackle-ai/knowledge) to support the "exports" field. If that is not possible, create a JSPM override to customize the exports field for this package.

Readme

@grackle-ai/knowledge

Knowledge graph subsystem for Grackle — keeps AI agent context (tasks, findings, sessions, workspaces) synchronized into a Neo4j knowledge graph with vector embeddings for semantic search.

This package is the Grackle-specific integration layer on top of @grackle-ai/knowledge-core. It re-exports the full core SDK (connection management, node/edge CRUD, ingestion pipeline, semantic search, graph traversal) and adds reference node synchronization — the glue that keeps the knowledge graph in sync with Grackle's relational database as entities are created, updated, and deleted.

Install

npm install @grackle-ai/knowledge

Requires Node.js >= 22 and a running Neo4j 5.x instance with vector index support. See the @grackle-ai/knowledge-core README for Neo4j setup instructions.

How It Works

Grackle entities (tasks, findings, sessions, workspaces) live in SQLite. To make them semantically searchable and graph-connected, this package provides reference node sync: when an entity changes, a corresponding reference node in Neo4j is created or updated with a fresh embedding vector. This lets agents query the knowledge graph with natural language and discover related context across entity types.

The package also includes pure text-derivation helpers that produce consistent, embeddable text representations from entity data — ensuring that the same task or finding always generates the same embedding input format.

Relationship to @grackle-ai/knowledge-core

Package Role
@grackle-ai/knowledge-core Domain-agnostic graph SDK — Neo4j client, embedders, chunkers, search, traversal. Standalone, no Grackle dependency.
@grackle-ai/knowledge Grackle integration layer — re-exports the core SDK and adds entity-aware sync, lookup, and text derivation.

If you are building on Grackle, use @grackle-ai/knowledge — it gives you the full core API plus Grackle-specific helpers in a single import. If you want a standalone knowledge graph library with no Grackle coupling, use @grackle-ai/knowledge-core directly.

License

MIT