Documentation

MetaLoom is built around two cooperating systems:

Loom manages assets, metadata, users, permissions and pipeline definitions.
Cortex processes media, extracts signals and feeds results back into Loom.

This page is the quickest way to orient yourself in the platform and jump to the right part of the documentation.

Getting Started

Run a full MetaLoom stack locally in minutes using the demo container. Covers Loom UI, Loom App and where to go next.

Demo Container Loom UI Container Images Chat & AI Agent Loom App

Task-oriented guides that string the whole platform together: deploying the stack with Docker or Kubernetes, building pipelines that transcribe, analyse scenes and translate content, and writing a processing node of your own.

Docker Deployment Kubernetes & Helm Helm Charts Transcription for Chat Scene Analysis Translation Custom Node (Python)

Loom

The media asset management server. Use these guides when you need the API surface, authentication model, configuration, artifacts or a feature overview.

REST API GraphQL API Java Client Authentication Configuration Chat & AI Agent Features Metrics Examples

Cortex

The media processing engine. Use these guides when you need node behavior, runtime configuration, deployment options or an overview of supported processing capabilities.

Pipeline Nodes Configuration Monitoring API Metrics Containers Custom Nodes (Java/Python)

Nodes

The built-in Cortex processing nodes — the atomic steps a pipeline is built from. Each node page covers what it does, the media it applies to, its input and output ports, requirements and configuration.

All Nodes Hashing & dedup Extraction & AI Media analysis Pipeline

Loom UI

The browser-based management interface bundled in the Loom server. Browse assets, chat with the AI agent, edit pipelines, and manage users, roles and API keys — all from the browser, with a screenshot tour of every area.

Overview & Screenshots Chat & AI Agent Asset Management Pipeline Editor Users, Roles & API Keys

Legal & Licensing

What you are allowed to run. MetaLoom is Apache-2.0, but the models its nodes load carry their own licenses — and a few of them forbid commercial use. Also covers how the source was produced and where AI code generation was involved.

Model Licenses Non-commercial models A clean commercial stack AI Disclosure

Start Here

If you are new to the platform, this is the shortest reading order:

Start with Getting Started to run a local demo in minutes.
Read Operation to understand the system boundary.
Read Pipeline Mechanism to understand how processing is modeled.
Continue into Loom or Cortex depending on whether you are integrating with the API or building processing workflows.
When you want to build something concrete — a deployment, a working pipeline — follow a Playbook.

Central Concepts

The MetaLoom platform revolves around a small set of concepts that appear throughout the documentation.

Asset

The main entity in Loom. An asset represents a file together with metadata, binaries, tags, reactions, comments and derived processing output.

See Loom features

Pipeline

A directed graph of processing nodes, authored in the UI and stored on Loom. Loom's engine runs the graph and delegates node tasks to Cortex workers.

See pipeline mechanism

Node

A single processing step such as hashing, OCR, thumbnail generation, face detection, captioning or synchronization back to Loom.

See pipeline nodes

Operation

Cortex runs as a worker daemon that registers with Loom, receives the node tasks Loom dispatches, and persists its results back over the REST API.

See how it operates

API and Client

Loom exposes a REST API, a Java client and the metaloom command line client for creating assets, authenticating, managing users and orchestrating processing workflows.

REST API · Java Client · Command Line

Deployment

Both systems ship as JVM artifacts and container images (loom-server, loom-demo, cortex-server, loom-session-runner), with Kubernetes deployment paths.

Container images · Helm charts · Loom artifacts · Cortex artifacts

Choose Your Path

Choose the documentation track that matches your task:

Goal	Start with	Then continue to
Try the platform quickly	Getting Started	REST API, Authentication
Integrate a client application with Loom	REST API or Java Client	Authentication, Configuration
Drive or script Loom from a terminal	Command Line	Pipeline Mechanism, Authentication
Understand how media gets processed	Operation	Pipeline Mechanism, Nodes
Deploy or operate the system	Docker Playbook, Kubernetes Playbook, Helm Charts	Monitoring, Loom Metrics, Cortex Metrics
Build a working pipeline end to end	Playbooks	Transcription, Scene Analysis, Translation, Custom Node
Learn the data model and feature surface	Loom Features	Nodes
Use or deploy the AI assistant	Chat & AI Agent	Coding Sandbox
Clear a deployment for commercial use	Model Licenses	Non-commercial models, AI Disclosure

Goal

Start with

Then continue to

Try the platform quickly

Getting Started

REST API, Authentication

Integrate a client application with Loom

REST API or Java Client

Authentication, Configuration

Drive or script Loom from a terminal

Command Line

Pipeline Mechanism, Authentication

Understand how media gets processed

Operation

Pipeline Mechanism, Nodes

Deploy or operate the system

Docker Playbook, Kubernetes Playbook, Helm Charts

Monitoring, Loom Metrics, Cortex Metrics

Build a working pipeline end to end

Playbooks

Transcription, Scene Analysis, Translation, Custom Node

Learn the data model and feature surface

Loom Features

Nodes

Use or deploy the AI assistant

Chat & AI Agent

Coding Sandbox

Clear a deployment for commercial use

Model Licenses

Non-commercial models, AI Disclosure

Architecture View

The overall system flow is:

Loom stores asset metadata, user state and pipeline definitions.
Cortex loads or receives work, processes media through pipeline nodes and emits results.
Results are stored locally, sent back to Loom, and exposed through APIs, events and search systems.

For the full runtime model, continue with Operation.