Overview

Copilot Memory has transitioned from an opt-in public preview to a default-enabled feature for GitHub Copilot Pro and Pro+ users. This enhancement allows Copilot agents to discover, store, and reuse knowledge about your codebase—including coding conventions, architectural patterns, and critical cross-file dependencies—across multiple tools and interactions.

How Copilot Memory Works

Repository-specific memories are strictly scoped to a single repository and validated against the current codebase before being used, ensuring stale or inaccurate context never influences suggestions. To maintain freshness, memories automatically expire after 28 days. Once enabled, Copilot Memory enhances three key surfaces:

• Copilot coding agent: Applies repository context when implementing tasks and opening pull requests
• Copilot code review: Uses learned patterns for more accurate, targeted review feedback
• Copilot CLI: Brings repository awareness to terminal workflows

Since memories are shared across agents, knowledge discovered by one tool immediately benefits others.

User Control and Management

No action is required to start using Copilot Memory—it's enabled by default for Pro and Pro+ users. However, individual users can opt out anytime in their personal Copilot settings under Features > Copilot Memory. Enterprise and organization admins retain full control over memory availability through Copilot policies at the enterprise and organization levels.

Repository owners can review and delete stored memories at any time in Repository Settings > Copilot > Memory, giving teams granular control over what gets remembered.

What This Means for Developers

This change aims to reduce context-switching overhead by enabling Copilot to maintain awareness of your project's unique patterns and conventions. Developers should expect more relevant suggestions that account for project-specific standards without needing to repeatedly explain context during conversations with Copilot agents.