LycheeMem FAQs

Name: LycheeMem
Author: LycheeMem

Question 1

How does LycheeMem organize its memory stores?

Accepted Answer

LycheeMem organizes memory into three complementary stores: Working Memory (for episodic context), Semantic Memory (a typed action store using SQLite and LanceDB), and Procedural Memory (a skill store for reusable 'how-to' knowledge).

Question 2

What is LycheeMem and its primary purpose?

Accepted Answer

LycheeMem is a compact, efficient, and extensible memory framework specifically designed for LLM agents. Its primary purpose is to provide sophisticated long-term memory solutions that enhance the capabilities of agentic AI systems.

Question 3

Is LycheeMem compatible with existing LLM agent frameworks?

Accepted Answer

Yes, LycheeMem offers seamless integration with frameworks like OpenClaw via a native plugin for persistent long-term memory. It also provides an HTTP MCP endpoint, allowing custom clients to connect easily.

Question 4

How does LycheeMem handle skills and procedural knowledge?

Accepted Answer

Procedural Memory functions as a Skill Store, preserving reusable 'how-to' knowledge as structured skill entries. It's enhanced by HyDE (Hypothetical Document Embeddings) retrieval, allowing the agent to find relevant skills even from vague user queries.

Question 5

What is 'Action-Aware Search Planning' in LycheeMem's Semantic Memory?

Accepted Answer

Action-Aware Search Planning is a core feature that analyzes user queries to drive intelligent, multi-channel retrieval (Full-Text Search, vector, tag, temporal) from Semantic Memory, optimizing for both factual answers and action execution.

LycheeMem

LycheeMem

Características Principales

Casos de Uso

Características Principales

Casos de Uso