Install Stackarr

Docker configuration

The supplied Compose file is the supported Stackarr installation. Put optional overrides in a .env file beside docker-compose.yml so updates do not overwrite them.

Common settings

TIMEZONE=Etc/UTC
STACKARR_BIND_IP=127.0.0.1
STACKARR_WEB_PORT=7777
MEDIA_ROOT=/srv/media
DOWNLOADS_ROOT=/srv/downloads
BACKUP_ROOT=/srv/backups

Create the host folders before starting services and make sure Docker can read and write them.

Why Stackarr uses the Docker socket

Stackarr needs /var/run/docker.sock to inspect and manage the containers selected during setup. This is powerful host-level access.

  • Keep the dashboard bound to localhost until authentication is configured.
  • Do not expose the local stdio MCP process over the network.
  • Use the observe or manage agent profile unless broader authority is intentional.
  • Review Agent Activity after automated changes.

Start and stop

docker compose --profile stackarr up -d app
docker compose stop app
docker compose logs -f app

Stackarr starts other services only after you select them during setup.

Choose an image channel

The moving early-access image is:

STACKARR_IMAGE=polyphonic/stackarr:alpha

Use a versioned tag when you want repeatable deployments:

STACKARR_IMAGE=polyphonic/stackarr:0.3.0-alpha.1

Connect an MCP client

Chat clients run on the Docker host. Generate the correct stdio configuration for the client you use:

docker exec app /app/bin/stackarr mcp config codex --profile manage

Run the generated command. Stackarr includes the required interactive stdio flag and selected authority profile. See MCP connections for every supported client, including private ChatGPT access.

Audit folder access

docker exec app /app/bin/stackarr permissions audit

The audit checks Stackarr's config, logs, downloads, backups, and the bind mounts used by enabled services.

On this page