smart-memory

What problem does it solve?

AI agents lose context between interactions, forcing users to repeat information and breaking workflow continuity.

Core Features & Use Cases

• Persistent storage of episodic, semantic, belief, and goal memories via a local FastAPI server.
• Semantic retrieval with Nomic embeddings and token‑bounded prompt composition.
• Hot working memory that tracks active projects, questions, and insights across sessions.
• Background cognition that generates reflections, consolidates memories, decays low‑importance items, and resolves belief conflicts.
• Tool interfaces (memory_search, memory_commit, memory_insights) for seamless integration with OpenClaw, LangChain, or custom agents.
• Session arc capture automatically creates summaries at checkpoints and session end.

Use cases include AI sales assistants recalling past client decisions, research bots maintaining project focus, and any autonomous agent needing continuous memory.

Quick Start

Request the smart‑memory skill to compose your agent’s initial context by providing your agent identity and a opening message.