Skip to main content

atmos ci cache

The atmos ci cache command group restores a well-known cache directory (the toolchain install path and anything else under the Atmos cache root) at the start of a CI step and saves it back at the end, using the active CI provider's cache store — the same store that actions/cache uses. This warm-starts the toolchain and other regenerable artifacts across CI jobs and workflow runs.

Experimental
Configure the CI Cache

Learn how to enable the cache, choose automatic vs. manual behavior, and customize the cache key, paths, and restore-keys in your atmos.yaml.

Lifecycle​

The cache lifecycle can run in a single Atmos invocation (automatic restore-on-start and save-on-end) or be spread across CI steps with the explicit subcommands:

Spread across steps
- run: atmos ci cache restore   # step 1: warm the cache
- run: atmos toolchain install  # step 2: use it (installs only what's missing)
- run: atmos ci cache save      # step 3: persist it for the next run

Both styles share one implementation. "Automatic" is just the same idempotent operations invoked by the process lifecycle, so manual and automatic invocations never double-execute (cache entries are write-once; an exact-key hit at restore time skips the save).

Requirements​

Saving and restoring content require running inside a supported CI provider (GitHub Actions today), which exposes the runtime cache credentials. Outside CI, these commands report that the cache is unavailable. The cache must also be enabled — see the configuration reference.

Subcommands​