Keep projects separate
Each workspace has its own entities, ontology, and changelog.
Getting started
Create a workspace, define what matters, and connect an agent. The console makes the same memory available for your team to inspect and manage.
In practice
You can try the workflow in the console before connecting an agent.
Built into the workflow
Each workspace has its own entities, ontology, and changelog.
Browse records, review held proposals, and follow recent activity from the console.
Create an access key and use MCP, the TypeScript SDK, or the HTTP API.
Observations from the console and connected clients pass through consolidation. Your ontology and policies apply to both.
Follow an observation through memoryCOMMON QUESTIONS
A workspace. Define a couple of classes or use the starter vocabulary, submit an observation, and open the resulting record. You can do all of it in the console before connecting any agent.
An isolated memory instance with its own entities, ontology, policy, and changelog. Keeping projects in separate workspaces keeps their vocabularies from colliding.
Yes. Observations from the console and from connected clients pass through the same consolidation path, with the same ontology checks and policies applied to both.
Once a small part of your world is modelled and you can see observations landing as expected. Then create an access key and connect through MCP, the TypeScript SDK, or the HTTP API.
Make your next update count
Create a workspace, define a type, and see your first observation become a traceable record.