GitHub Enterprise Server 3.23 release candidate GitHub released the GitHub Enterprise Server 3.23 release candidate, adding AI secret detection that runs a self-contained, fine-tuned model on the appliance's CPU with no GPU, hosted LLM, or outbound connection, including in air-gapped environments; the feature is in public preview and subject to change. The release also extends the GitHub Advisory Database to flag known malicious packages in PyPI, Maven, RubyGems, NuGet, Go, Rust, and Composer in addition to npm, makes Dependabot wait until a package version has been on its registry for at least three days before opening an update pull request, and adds opt-in OAuth access tokens that expire after eight hours with refresh tokens. Other changes include logical OR, explicit AND, and parentheses grouping in GitHub Projects filters, and the ability to move the Redis primary to a dedicated node in a GHES cluster. GitHub Enterprise Server 3.23 release candidate GitHub Enterprise Server GHES 3.23 is now available and introduces new capabilities across the platform. Here are a few highlights in the 3.23 release: - When working in GitHub Projects, you can create more precise project views by combining filters with logical OR , explicit AND , and parentheses for grouping. For example, status:Done OR status:Todo is:issue shows issues with either status. - If your enterprise uses GitHub Secret Protection, you can detect unstructured secrets entirely within your network using AI secret detection on GHES, including in air-gapped environments. A self-contained, fine-tuned model runs on the appliance’s CPU without a GPU, hosted LLM, or outbound connection. This feature is in public preview and subject to change. - As a developer, you can identify known malicious packages in PyPI, Maven, RubyGems, NuGet, Go, Rust, and Composer through the GitHub Advisory Database, in addition to npm. The database automatically ingests malware advisories from OpenSSF’s malicious-packages repository to expand coverage across these ecosystems. - By default, Dependabot waits until a package version has been available on its registry for at least three days before opening a version update pull request. This cooldown gives maintainers and the community more time to detect and report compromised or buggy releases before users merge updates. - As an OAuth app developer, you can limit token lifetimes by using access tokens that expire after eight hours and refresh tokens to obtain new access tokens. Token expiration is opt-in for existing OAuth apps and the default for newly created apps. You can request expiring tokens with scope=offline access during authorization for testing and gradual rollout, or require expiration in your app registration. Before requiring expiration, you must update all clients to request, store, and use refresh tokens to avoid breaking sign-in for clients that do not support them. This feature is in public preview and subject to change. - In your GHES cluster, you can now move the Redis primary from a node that also runs other services, such as MySQL, to a dedicated node. Redis then uses the dedicated node’s resources and data volume rather than sharing them with those services. This procedure does not attach a new disk to an existing node or use an external Redis service. Release candidates let you try the latest features early and share feedback before general availability. To learn more, see About upgrades to new releases https://docs.github.com/enterprise-server@3.23/admin/overview/about-upgrades-to-new-releases and the GHES 3.23 release notes https://docs.github.com/enterprise-server@3.23/admin/release-notes . To get started, download the GHES 3.23 release candidate https://enterprise.github.com/releases/3.23.0/download . If you have feedback or questions about the release candidate, contact GitHub Support https://support.github.com/features/enterprise-administrators-server .