GitHub Intelligence Introduces Cross-Session Memory and Enhanced Code Search

GitHub has announced GitHub intelligence, a new context and learning layer designed to help AI agents understand team preferences, repository conventions, and development history across projects. According to the announcement, the feature set aims to prevent agents from treating every session as a fresh start by leveraging data graphs from GitHub.

Core Capabilities

• GitHub Copilot Memory: Generally available on paid Copilot plans, allowing Copilot to remember personal preferences and repository conventions across sessions and clients such as VS Code, Slack, and the Copilot CLI.
• Copilot Dreaming: Currently in public preview, this feature reflects on past work to curate personal memories, manage repository customs, and convert accepted code-review feedback into ongoing guidance.
• GitHub Code Search: Enables Copilot to search exact text, symbols, patterns, and semantic descriptions across related projects and dependencies without requiring a local clone.

Practical Implications and Verification

Teams can utilize these tools to reduce the repetition of instructions and streamline complex workflows like bug investigations and cross-repository migrations. Users can verify availability, setup prerequisites, and policy controls by checking official GitHub documentation or managing settings directly within their Copilot account dashboard.