Memora — постоянная память для ИИ-агентов

★ 523

Memora is a memory layer for AI agents over the MCP protocol: it gives agents persistent collective memory with structured storage, semantic retrieval, graph relations, and source-backed cross-session context. Reach for it when you want an agent to "remember" the results of its work and retrieve them later: store observations, TODOs, relations, and documents, then pull relevant records by topic (memory_digest) along with related graph edges and source IDs. It keeps data in SQLite with optional cloud sync (S3, R2, D1), supports hierarchical organization, LLM-based dedup, linking, and semantic search, and offers a live graph and integrations. Its focus is a reusable memory store with graph and semantics for agents via MCP, not the agent itself or a general-purpose vector database: it is a component you attach to an agent so context outlives a single session.