Show HN: Spomory – graph-based MCP memory with an RL-trained forgetting policy

Category: ai-ml

Tags: mcp, knowledge-graph, memory, reinforcement-learning, llm

Score: 5.3/10 (Innovation: 6, Technical: 5, Documentation: 3, Utility: 5)

Spomory is a graph-based memory system for MCP (Model Context Protocol) that extracts structured (subject, predicate, object) triples from text via an LLM and manages itself with an RL-trained forgetting policy. It is interesting for combining knowledge-graph memory with a learned retention/forgetting mechanism, aiming at explainable and self-managing agent memory.

Target audience: AI engineers and agent developers

Repository: https://spomory.yliuai.com/

View on Hacker News