Architecture
The DKG Node is built as a modular project with two core runtimes:
the DKG Engine, which powers network communication and implements the core protocol
the DKG Node Runtime, which hosts an AI Agent with MCP capabilities
Adding functionality is done through Plugins, which is where you'll likely spend the majority of your time coding. Conceptually the architecture is illustrated below.

DKG Node server (MCP) and Agent
This is the “brain” of your node. It runs all plugins and connects to the underlying services of the DKG node, that AI agents or other applications use. It allows agents to publish, query, and retrieve Knowledge Assets directly from the OriginTrail DKG. It can also expose REST APIs (via Express) so your apps can interact with the node over HTTP.
Plugins
Plugins are like mini-apps for your DKG Node AI Agent - small add-ons that unlock new functionality. They can provide MCP tools (for AI agents), HTTP endpoints, or both.
Some useful built-in plugins include:
DKG Essential Plugin - includes the basic tools for publishing and retrieving knowledge.
OAuth 2.1 authentication - controls who can access your node.
Swagger - automatically documents available APIs.
DKG Node engine
The DKG engine (formerly known as ot-node) implements the core OriginTrail protocol and is considered a dependency (not intended for implementing agent functionality). The DKG Node engine should be kept up to date in order to maintain reliable and efficient communication with the rest of the network. It implements blockchain interactions, takes care of loading new knowledge assets from the network and their validation, performs staking transactions (in case of Core nodes) etc.
Last updated
Was this helpful?