A developer from the Engrava project outlines five failure modes to check agent memory stores against, including hallucinated metadata, stale facts, and missing conflict resolution. The post highlight…
Engrava 0.6.0 scored 81.6% micro on the LongMemEval-S benchmark, a slight drop from the 0.5.0 release's 82.4%, with both runs using identical readers, judges, and scorers. The company emphasizes repro…
Engrava 0.5.0 splits its MCP server into a standalone package, engrava-mcp, decoupling it from the core library. The server is now listed in the official MCP Registry as ai.sovantica/engrava, and the …
A developer built Engrava, a typed knowledge graph with hybrid search in a single SQLite file, designed to give long-running AI agents deterministic, inspectable memory. The library uses a consolidati…