Skip to main content

Upgrade

driverforge keeps itself, the Anvil Agent, and your embedded SDK current. Here's how the pieces fit.

Update notices​

After a command runs, the CLI checks (at most once a day, cached) whether a newer driverforge release is available, and prints a one-line notice if so. It's quiet by design: the notice goes to stderr, never stdout, and never appears in CI or non-interactive shells. Turn it off with --no-update-check.

Upgrading the CLI​

driverforge is installed by a package manager, which owns the binary, so you upgrade it the same way you installed it:

brew upgrade driverforge      # macOS
scoop update driverforge      # Windows

# Linux / WSL: re-run the install script
sh -c "$(curl -fsSL https://go.driverforge.com/get)"

driverforge upgrade shows everything upgradable in one place: the CLI (advisory: it prints the exact command for how you installed it), the agent, and the embedded SDK. It applies them interactively.

Upgrading the Anvil Agent​

The driverforge upgrade picker is how agents get upgraded: select the Anvil Agent row and confirm, and the CLI updates the agent on the selected controller to the latest release. Upgrades are interactive-only: non-interactively, driverforge upgrade reports and never mutates.

Upgrading the SDK​

The Driverforge SDK is embedded in your driver, and the SDK row of the driverforge upgrade picker is the only way to update it. It re-embeds the latest release; a clean git tree is required.