This page lists what the local Docker Compose deployment is for and what you need before following Quickstart.
When to use this deployment
Use the Compose stack to evaluate Kumoss, develop on it, run demos, or test changes on a workstation or a throwaway virtual machine that only trusted people can reach.
Do not use it for shared or production environments. As shipped it has no
TLS, well-known default credentials for PostgreSQL and RustFS, an
unauthenticated Phoenix console, and, unless you enable OIDC, no
authentication at all: every request runs as a built-in local identity
that holds the devops operation role and the panel admin role. It
publishes ports 80 and 9000 on every interface of the host. For anything
shared, read
Deploy to production.
Requirements
-
Credentials for one LLM provider supported by LiteLLM. See LLM providers and models.
-
Cloud credentials for the cloud your OpenTofu/Terraform code targets: the variables your providers read to create and change the resources
planandapplymanage. They are given to the IaC sidecar. See Terraform providers. -
A personal access token for your Git provider, with permission to push branches and open pull requests.
-
Git, and Docker Engine with Docker Compose v2.24 or later.
-
As a rough guideline, at least 8 GB RAM and 10 GB free disk.
Separate state-backend credentials are optional. By default Kumoss
manages Terraform state itself in the bundled object store, so the cloud
credentials above are all the IaC sidecar needs. A second grant is
required only if you set storage.terraform_state_bucket to "" and let
each target repository declare its own backend, in which case it must
reach the store that repository’s terraform { backend … } block names.
See
Configure state backends.