Copilot Memory Store

Example Use Cases

• 1

  Store and recall project-specific knowledge and preferences.

• 2

  Compress and inject relevant context within LLM token limits.

• 3

  Interact naturally with memory tools via VS Code Copilot.

Description

Copilot Memory Store provides a local JSON-based memory system designed to enhance context engineering for large language models by storing, searching, and compressing relevant memories. It features an interactive CLI, an MCP stdio server exposing tools and prompts, and a custom VS Code Memory agent for natural language interaction. The server supports context compression with optional LLM summarization via DeepSeek, enabling efficient context injection within character budgets to improve AI coding assistance.