From the source
GitHub is rolling out GitHub intelligence, a context and learning layer built from the graph of GitHub data, so Copilot agents can carry team knowledge between sessions instead of starting fresh.
- Copilot Memory is now generally available on paid Copilot plans, letting Copilot remember personal preferences and repository conventions across sessions and clients, including the Copilot app, Copilot CLI, VS Code, Copilot cloud agent, Copilot code review, Slack, Teams and GitHub Mobile.
- Among Copilot CLI sessions reviewed during the public preview where the agent used memory, GitHub says 93% showed better adherence to developer preferences and 42% showed more efficient workflows attributed to the memories included in the session context.
- Copilot dreaming, now in public preview, reflects on past work to keep that context current through three dream types: personal memory curation, repository customization, and code review customization.
- Dreaming is opt-in, can add, update and remove personal and repository memories, and proposes changes to checked-in skills and instructions as pull requests for team review; dream runs consume AI credits.
- GitHub code search lets Copilot find code, configuration, schemas and documentation across related projects and dependencies without a local clone, supporting exact text and symbol search, regular expressions, and semantic search by what code does.
- Copilot can narrow results by repository, path or language, and the /research command lets it inspect implementations and callers across repositories for tasks such as planning a migration or investigating a bug.
