Getting Started

This path takes a new install of NEXUS all the way to using every major surface once: signing in, creating an agent, running a session, arranging your panes, reading the Neural Map, watching the Nexus Room, sending a mesh message, pinning durable memory, and connecting an external CLI over MCP. Following every page in order takes about 30 minutes. Each page ends with a Next link, so you can stop and resume anywhere.

Work through the pages in order:

1. What is NEXUS? — the problem NEXUS solves and where the developer-facing surface sits.
2. Install and first launch — get NEXUS running and past sign-in for the first time.
3. The workspace tour — the left rail and every mode it switches between.
4. Create your first agent — the three ways to create one, persona, and autonomy modes.
5. Run your first session — start a shell as that agent and read its live status.
6. Panes and layout — splitting, grids, Focus Mode, and switching workspaces.
7. The Neural Map — your workspace as a graph you can navigate.
8. The Nexus Room — watch agents work live, and start a team.
9. Agent-to-agent messaging — mesh messages between two agents, and where they queue.
10. Pinned memory — pin a durable project rule and review what's pinned.
11. Connect a CLI over MCP — register a hosted CLI agent as an MCP client of NEXUS's local server.
12. First tool call — send a real request and read a real response back.
13. Where to go next — a map of the Guides, Reference, and Explanation sections for everything this path only introduces.

If you already know what NEXUS is and only need the protocol contract, skip ahead to the MCP protocol reference and the MCP tools reference.