Persistent memory for AI agents — REST API, MCP, OAuth, CLI, dashboard. Semantic search, knowledge graph, automatic consolidation. One self-hosted service, every transport.
Native Milvus search_memories with server-side filter pushdown, hooks JSONL transcript fix, and dep updates. ~1,968 tests.
Milvus backend now overrides search_memories, retrieve_with_quality_boost, and recall_memory. Filters (tag, type, date) are pushed to the Milvus server inside the ANN call — no more N round-trip fallbacks to the base class. Completes the medium-priority method set from issue #888.
Claude Code transcript format changed to JSONL with a message wrapper. The auto-capture hook now correctly unwraps each line before extracting tool calls, restoring transcript-based memory capture for current Claude Code versions.
retrieve_with_quality_boost re-ranks results by blending embedding similarity with stored quality scores server-side. recall_memory adds temporal recency weighting. Both use Milvus native ANN search for full performance.
Claude.ai connects to your memory server through Streamable HTTP with OAuth 2.1 authentication.
Battle-tested with comprehensive testing and optimized for performance.
Install from PyPI and connect your AI agents to persistent memory.