Container Images

MetaLoom ships as a small set of OCI (Docker/Podman) container images. All images are published under the metaloom/ namespace.

Image Overview

Image Purpose Built from

metaloom/loom-server

The Loom backend server (REST/gRPC API, DB access, auth, UI). Production image.

loom/containers/server

metaloom/loom-demo

All-in-one demo: Loom server pre-seeded with a database and sample assets. Best for a first run.

loom/containers/demo

metaloom/cortex-server

The Cortex processing daemon (media analysis worker). Registers with Loom and serves nodes.

cortex/container

metaloom/loom-session-runner

The isolated per-chat coding sandbox spawned by the Loom AI agent. Not started manually — Loom provisions it on demand. See Chat & AI Agent — Coding Sandbox.

loom/agent/session-runner

Both loom-server and loom-demo are available in two variants:

  • JVM (default tag, e.g. :latest) — Eclipse Temurin JRE running the shaded JAR.

  • Native (-native suffix, e.g. :latest-native) — a GraalVM ahead-of-time compiled binary with faster startup and lower memory.

Default Ports

Service Port Notes

Loom REST API + WebSocket + UI

8092

The single port clients, the UI and Cortex workers all connect to. Env: LOOM_SERVER_REST_PORT

Loom gRPC

8091

Optional gRPC surface. Env: LOOM_SERVER_GRPC_PORT

Loom monitoring

8989

Health / metrics. Env: LOOM_SERVER_MON_PORT

Cortex monitoring

8093

Cortex health / readiness (/api/health, /api/ready). Env: CORTEX_MONITORING_PORT

Session Runner

8080

Internal only — reached by Loom over the pod/loopback network, never exposed publicly.
Note
 Cortex connects out to Loom on port 8092; Loom never dials in to Cortex. Older documentation referenced a Cortex↔Loom port of 7733 — the current default is 8092.

See Getting Started to run the demo image, and Loom Containers / Cortex Containers for production deployment examples.