An increasing number of malware attacks exploit systems at package installation time, harvesting credentials on CI/CD hosts or developer workstations without even running the package’s code. To stem this tide of attacks, GitHub is introducing a new package firewall capability. The package firewall sits between endpoints like Actions runners or coding agents and the upstream package repositories, inspecting both metadata requests and package downloads for signs of malicious activity. The package firewall is available today in public preview for all GitHub Enterprise Cloud customers, including data-resident GitHub Enterprise Cloud.

First, administrators set policy controls for their enterprise. The controls available today include a configurable cooldown period, checking the GitHub Advisory Database for malware or vulnerability alerts, and supply-chain checks like attestation downgrades.

Then endpoints like coding agents and CI/CD hosts need configuration to use the firewall for their requests instead of pointing directly at the upstream package registries. Metadata requests for package lists as well as the actual package downloads then flow through the firewall, which evaluates them against the policy. If a package is blocked erroneously, the user can request an exception with a workflow similar to other GitHub Advanced Security features.

The package firewall is available in public preview as we scale up the infrastructure behind it and continue to add features based on customer demand. It is available to GitHub Enterprise Cloud customers, including GitHub Enterprise Cloud with data residency. The service is free while in the preview period. When it reaches general availability, it will be charged based on metered usage, similar to Actions billing.

Currently the package firewall only supports npm packages.

For detailed setup instructions, see Protecting endpoints with package firewall. Join the discussion in GitHub Community.