Back to rankings

GitHub Copilot Memory: Now Generally Available on Paid Plans

GitHub CopilotView original changelog

GitHub made Copilot Memory generally available on paid Copilot plans as part of GitHub intelligence, its context and learning layer for agents. Copilot can now remember personal preferences and repository conventions across sessions and clients, including the Copilot app, Copilot CLI, VS Code, the cloud agent, code review, Slack, Teams and GitHub Mobile. In GitHub's review of public-preview CLI sessions that used memory, 93% showed better adherence to developer preferences and 42% showed more efficient workflows.

Key Takeaways

  • Copilot Memory is now generally available on paid Copilot plans, ending its public preview that began in early 2026.
  • Memories carry personal preferences and repository conventions across sessions, so developers stop repeating the same instructions.
  • One memory store works across clients: the Copilot app, CLI, VS Code, cloud agent, code review, Slack, Teams and GitHub Mobile.
  • In preview CLI sessions that used memory, GitHub reported 93% better adherence to preferences and 42% more efficient workflows.
  • Copilot generates and curates memories on its own, and users can view or delete saved personal preferences in Copilot settings.
  • Memory is one pillar of GitHub intelligence, alongside Dreaming for keeping context current and code search for reaching beyond the local checkout.

Copilot Memory leaves preview

GitHub Copilot Memory is now generally available on paid Copilot plans. It is the first of three capabilities GitHub groups under GitHub intelligence, described as the context and learning layer sourced from the graph of GitHub data. The goal is for agents to build on what a developer and their team already know, instead of treating every session as a fresh start.

What Copilot can remember

Copilot can retain two kinds of knowledge across sessions and clients. The first is personal preferences. A developer can say "Remember to start my pull request descriptions with a TL;DR", and Copilot can reuse that preference in every Copilot experience. The second is repository conventions, such as the right test command or how a team expects Copilot to check its work. Capturing these means developers spend less time repeating the same guidance at the start of each session.

Memories can be created in the Copilot app, Copilot CLI, VS Code, the Copilot cloud agent, Copilot code review, Slack, Teams and GitHub Mobile. Copilot sessions also generate and curate memories on their own, learning from each interaction. A repository memory can record why an edge case matters and which test covers it, and GitHub intelligence lets agents use that context with controls over what they can access.

Measured impact from the preview

GitHub reported results from Copilot CLI sessions reviewed during the public preview in which the agent used memory. In those sessions, 93% showed better adherence to developer preferences and 42% showed more efficient workflows, which GitHub attributed to the memories included in the session context.

Managing memories

Developers can interact with memories directly in the Copilot app, Copilot CLI and VS Code. Personal memories can also be managed in Copilot settings, where saved preferences can be viewed or deleted. Memory pairs with Dreaming, which curates and updates memories over time, and with cross-repository code search, which supplies context from related projects.

GitHub Copilot Memory Now Generally Available | Yet Another Changelog