Skip to content

Changelog

For the full changelog, see CHANGELOG.md in the repository.

[0.2.0] — 2026-04-03

Major feature release: Neo4j persistent backend, scientific entity schema, bulk document ingestion, Qdrant vector search bridge, and cross-session entity deduplication. See CHANGELOG.md for full details.

[0.1.0] — 2026-02-17

First public release. NeuroWeave is an async Python library that transforms AI conversations into a live knowledge graph.

Highlights

  • NeuroWeave facade class — the single entry point with process(), query(), and get_context() methods.
  • LLM-powered extraction — entities and relations extracted from conversational messages via Claude or mock LLM.
  • Structured and NL queries — query by entity names and filters, or ask natural language questions.
  • Event subscription — async pub/sub for real-time graph mutation notifications.
  • Cytoscape.js visualization — optional browser-based graph viewer with WebSocket live updates.
  • ~308 tests across 16 test files with full integration coverage.
  • Typed — full type annotations and PEP 561 py.typed marker.

Dependencies

Python 3.11+, anthropic ≥0.42, networkx ≥3.2, fastapi ≥0.115, structlog ≥25.5.