Memory FAQs

Question 1

What is Memory (mcp-memory)?

Accepted Answer

Memory is an AI tool providing persistent, structured, and episodic memory with semantic search capabilities for AI clients, leveraging local embeddings and LanceDB for efficient, local data management.

Question 2

What types of memory does Memory support?

Accepted Answer

It supports a structured memory layer for agent identity, user profiles, project states, and feedback, alongside an episodic memory layer through indexed conversation chatlogs.

Question 3

How does Memory handle embeddings and vector storage?

Accepted Answer

Memory uses Ollama's `nomic-embed-text` model for local embeddings and LanceDB as its local vector database, enabling persistent storage and incremental reindexing of data.

Question 4

Is Memory compatible with various AI clients?

Accepted Answer

Yes, Memory functions as a local MCP server designed to be compatible with any stdio-based AI client, offering flexible integration into your AI workflows.

Question 5

What content can I semantically search using Memory?

Accepted Answer

You can conduct semantic searches across your structured memories, conversation history (chatlogs), codebase files (TypeScript/JavaScript), and documentation files (Markdown, SQL, JSON).

Memory

Características Principales

Casos de Uso

Memory

Características Principales

Casos de Uso