Build & integrate

Follow this path if you will run, extend or integrate the Signals and Aggregator DPGs. It builds the shared vocabulary first, then goes deep on the architecture and data flows, and ends hands-on: a local stack, a running Signals DPG, and the API reference to build against.

1

Introduction

The minimum shared context — what the Blue Dots Economy is and why it exists.

2

Signals (Blue Dots)

The core object model in plain language — what a signal is and how it flows through the network.

3

Participant Profiles

The three participant types and the shape every profile follows.

4

High-Level Architecture

The two verticals plus shared datastores — apps, auth, APIs and integrations in one layered view.

5

Schema-Driven Model

The design idea everything hangs on — schemas define networks, domains and items so domains evolve without code changes.

6

Data Model

Entities and the ER diagram — how items, schemas and partitions are modelled across both datastores.

7

Read & Write Paths

How data actually flows — the two read layers, count-first discovery, caching, and the controlled write paths into Signals.

8

Local Stack (Docker)

Hands on — bring up Postgres, Redis, Keycloak and Mailpit for local development.

9

Signals DPG Setup

Stand up the first DPG — install, migrate and run the Signals backend and UI.

10

API Conventions

Build against it — the Signals and Aggregator endpoints, auth headers and error conventions.

Go deeper

Identity & AuthOne shared Keycloak realm across the stack — OIDC logins for people, client-credentials tokens for integrating DPGs.

Tech StackRuntimes, frameworks, datastores and tooling across the Signals and Aggregator DPGs.

CI/CD & Build PipelineHow code becomes a running deployment — image builds, GHCR, image pinning, and Helm-based delivery.

ConfigurationHow configuration-as-code works across the Blue Dots DPGs, and where each value lives.

Building a conversational agent? Start with the AI Diffusion DPG architecture, then run it locally with AI Diffusion DPG Setup and talk to it through the bridge, its default OpenAI-compatible endpoint. Web chat (local/dev only), the CLI, voice and MCP are optional channels.

Prefer browsing by role? See Who Is This For.