Update and uninstall

Upgrade the server and its runtime integrations, check what the CLI has installed, and remove Lore cleanly. You get the release channels, the backup to take before an upgrade, and a list of what an uninstall leaves on disk.

UpdateCopy Anchor Link

Update the server and its runtime integrations with one command:

npx @loremem/cli update

What update does depends on how the server is run:

• Docker-managed server (docker_managed is true in ~/.lore/config.json): downloads the current docker-compose.yml and keeps the previous file as docker-compose.yml.bak.<timestamp>. It sets the image tag for the release channel, runs docker compose pull and docker compose up -d, and waits for the health check.
• Any other server: a Loremem Cloud server, an existing server, or a Compose stack you manage. Docker and the server aren't touched. For your own stack, run the commands in the next section.
• Runtime integrations: update refreshes each runtime that is installed or partial. Use --channels to choose the runtimes yourself.
• Failures: update exits with code 1 rather than reporting success. That happens when the release can't be resolved, a selected runtime fails, or a selected runtime is skipped, for example because its CLI is no longer installed. For a runtime you removed, run uninstall --channels <runtime>, or leave it out with --channels.

Release channelsCopy Anchor Link

Update a manual Compose stackCopy Anchor Link

From your Compose directory, pull the new images and recreate the containers:

cd ~/lore-server
docker compose pull
docker compose up -d
curl -fsS http://127.0.0.1:18901/api/health

Check the installCopy Anchor Link

npx @loremem/cli status

The output shows the config path, the base URL, the installed version, whether the CLI manages Docker, and whether a token is saved. It never prints the token. It also lists the detected runtimes and each integration's state: installed, partial, missing, or unknown. The command doesn't contact the server.

Back up before you upgradeCopy Anchor Link

1. Confirm that backups run. backup.enabled and backup.local.enabled should be true in the settings (see the settings reference). Check that SNAPSHOT_DATA_DIR holds a recent backup.

2. Dump the database from your Compose directory:

   Copy Text

   docker compose exec -T postgres sh -c 'pg_dump -U "$POSTGRES_USER" "$POSTGRES_DB"' > lore-$(date +%Y%m%d).sql

3. Copy .env and ~/.lore/config.json somewhere safe. They hold the database password and the token.

UninstallCopy Anchor Link

uninstall removes runtime integrations. On a terminal, it lists the runtimes and asks Uninstall these channels? [y/N]. Without a terminal, it doesn't ask.

npx @loremem/cli uninstall                          # all runtimes, with a prompt
npx @loremem/cli uninstall --channels claudecode,pi # only these runtimes
npx @loremem/cli uninstall --purge -y               # also remove config and Docker data

Stop the stack before you purgeCopy Anchor Link

uninstall doesn't stop Docker. If the CLI installed the local stack, stop it before you purge:

cd ~/.lore/docker && docker compose down
npx @loremem/cli uninstall --purge -y

docker compose down stops the containers and keeps the data in place. The purge then deletes that data.

What uninstall leaves behindCopy Anchor Link

Uninstall removes integrations that the CLI can reach. It doesn't remove your own Compose directory, your data, or other files you wrote. Delete those yourself when you no longer need them.

Legacy shell installersCopy Anchor Link

The shell scripts scripts/install.sh and scripts/uninstall.sh in the repository are frozen. They are kept for compatibility and get no new features. Use npx @loremem/cli for installs, updates, and uninstalls. Run npx @loremem/cli status first to see what the CLI detects on a machine that an older script installed.