kael-bit/engram-rs

Overview

A memory engine for AI agents that organizes information along a time axis with three-layer decay and promotion and a space axis with a self-organizing topic tree. It provides hybrid search and LLM consolidation in a single Rust binary.

Best for

Best for
Developers building AI agents that need persistent, structured memory

Use cases

• Persisting agent memory across sessions with automatic decay
• Organizing knowledge into a self-organizing topic tree
• Retrieving relevant context using hybrid search

Notes

A memory engine for AI agents that organizes information along a time axis with three-layer decay and promotion and a space axis with a self-organizing topic tree. It provides hybrid search and LLM consolidation in a single Rust binary.

25 stars on GitHub. Last updated 2026-03-03. Licensed MIT.

Use cases

• Persisting agent memory across sessions with automatic decay
• Organizing knowledge into a self-organizing topic tree
• Retrieving relevant context using hybrid search

Pros

• Single Rust binary simplifies deployment
• Combines temporal decay and topic organization for structured memory
• Hybrid search enables flexible retrieval of relevant information

Cons

• Low star count indicates early development stage
• Rust dependency may require compilation for some environments
• Limited documentation or examples beyond the repository

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.