Vector Memory FAQs

Question 1

Is my data private and secure with Vector Memory?

Accepted Answer

Yes, absolutely. All your session data and generated embeddings remain entirely local on your machine. Vector Memory processes information from Copilot's local database and stores its index locally. Nothing is sent to any external service, ensuring your privacy.

Question 2

Does Vector Memory require a powerful computer or GPU?

Accepted Answer

No, Vector Memory is designed for efficiency and runs comfortably on any laptop CPU. The ONNX embedding model is tiny (~34 MB) and inference is fast, requiring no GPU. The server also automatically shuts down after 5 minutes of inactivity to conserve resources.

Question 3

What is Vector Memory and how does it help GitHub Copilot CLI?

Accepted Answer

Vector Memory is an MCP server that integrates local semantic vector search to give GitHub Copilot CLI persistent long-term memory. It allows Copilot to recall past conversations, code changes, and decisions across all sessions by meaning, rather than just keywords, significantly enhancing its contextual understanding.

Question 4

How do I install Vector Memory for GitHub Copilot CLI?

Accepted Answer

Installation is straightforward. You only need Node.js (v18+) installed. Then, locate or create your Copilot CLI `mcp-config.json` file and add the provided `vector-memory` server entry. Restart Copilot CLI, and the server will automatically download, install, and launch itself.

Question 5

What kind of information can Copilot CLI recall with Vector Memory?

Accepted Answer

With Vector Memory, Copilot CLI gains the ability to semantically search across all past session history. It can recall entire conversations, specific code changes, decisions made, or even topics discussed by their meaning, providing richer context for current tasks.

Vector Memory

Vector Memory

Key Features

Use Cases

Key Features

Use Cases