Show HN: IHMT – Infinite Hierarchical Memory Tree for AI Agents

Category: ai-ml

Tags: ai-agents, memory, mcp, local-first, hierarchical-retrieval

Score: 7.0/10 (Innovation: 6, Technical: 7, Documentation: 8, Utility: 7)

IHMT is a local, file-based long-term memory system for AI coding agents, using a hierarchical tree of text summaries instead of vector embeddings to enable fast, token-efficient retrieval across sessions. It aims to solve context persistence and stale fact handling with explicit time-awareness and cross-agent sharing via MCP.

Target audience: AI developers and engineers using coding agents (e.g., Claude Code, Codex) who need persistent, shareable memory across sessions and models.

Repository: https://github.com/gonzaroman/IHMT-MEMORY · Python · MIT · 38 stars

View on Hacker News