Offline CI - Jolter

Offline CI works when Jolter can reuse complete local installations or cached metadata and verified artifacts.

When to use offline mode

Use JOLTER_OFFLINE=1 for:

isolated runners with prewarmed caches;
release pipelines that must not contact providers;
disaster-recovery tests;
corporate environments where downloads happen through a separate approved job.

Do not enable offline mode until the cache contains everything the job needs.

Prepare the cache

On a connected runner with the same operating system and architecture:

jolter sync
jolter doctor
jolter cache status

Preserve:

$JOLTER_HOME/cache

Exact versions simplify offline operation:

jolter.json

{
  "runtime": {
    "node": "24.5.0"
  },
  "tools": {
    "pnpm": "10.12.1"
  }
}

Broad selectors such as latest and lts require cached provider metadata.

Run offline

export JOLTER_OFFLINE=1
jolter setup-ci --no-progress

PowerShell:

$env:JOLTER_OFFLINE = "1"
jolter setup-ci --no-progress

Cache misses

If offline setup fails:

jolter doctor --no-progress
jolter cache status

Check:

JOLTER_OFFLINE is set intentionally;
the cache was restored to the same JOLTER_HOME;
the cache was built for the same OS and architecture;
broad selectors have cached metadata;
jolter cache clean was not run before the offline job.

Security note

Jolter re-verifies artifact integrity before publication, but writable cache metadata and already-installed executables remain local filesystem state. Do not share writable caches across untrusted repositories or branches.

​When to use offline mode

​Prepare the cache

​Run offline

​Cache misses

​Security note

When to use offline mode

Prepare the cache

Run offline

Cache misses

Security note