Agents > Xedant MultiAgent > Docs > Getting Started

Getting Started

Xedant MultiAgent is the central control panel for a fleet of AI workers — instances of Xedant Agent. It connects to any number of Linux servers over SSH (SSH is the standard way to administer a remote server securely), deploys agents and products on them (Research, Analytics, Data, Proxy, MultiAgent itself, plus chatbots, Telegram, UI and CRM), collects their chats and analytics into one interface, and gives every app its own address on the server through Nginx routing (Nginx is the web server that hands out those addresses).

System requirements: MultiAgent itself is one Docker container (Docker Setup) or the Windows installer (Windows Installation). The servers it manages are Linux (Ubuntu/Debian) with SSH access and root/sudo rights. MultiAgent installs Docker and everything else on those servers by itself.

First Login

Once the container is running, open http://localhost:3990. There are no login and password environment variables to configure — the first user is created with the “Register” button on the login page, and later accounts the same way. Authentication uses JWT tokens valid for 90 days, so you will rarely need to sign in again.

Main Screen

After signing in you land on the main screen with navigation cards for every section of the product. The permanent navigation tree lives in the left sidebar: one root row per section (Servers, File Manager, Agents, Tools, Workspaces, Accounts, Analytics, Models, Skills, Licenses) with live sub-rows for real things — every server and agent you add appears in the tree on its own. Right-clicking a row opens an action menu: on a server row it is “Add Agent”; on an agent or app row it is Stop, Restart, Recreate. The status next to the row updates right there, without leaving the page.

  • Servers — connect and onboard managed servers, Nginx routing and certificates;
  • Agents — creating, deploying and the whole life cycle of agents and products;
  • Tools — a library of Docker Compose service templates;
  • Workspaces — grids of live agent chats;
  • Accounts — LLM provider accounts, quotas and balances;
  • Analytics — centralized statistics across all agents;
  • Models — a catalog of model configurations: create a model and push it to the online agents you pick — the agent’s record with the same name is replaced, and private variables rebuild the agent’s container. Importing models from an agent works the same way: when a name matches, the catalog record is replaced and no duplicate is created. A model name may contain any characters except /, ? and #;
  • Skills — a catalog of skills: a name, a color and a set of text files with a required SKILL.md. A skill is pushed to the online agents you pick (the agent’s folder of the same name is replaced exactly; containers are not rebuilt) or imported from an agent (a matching name replaces the catalog record);
  • Licenses — license keys of agents.

When the AGENT_API_URL and AGENT_API_KEY environment variables are set, an Own Agent panel appears at the bottom of the sidebar: a list of active chats with timers and token/cost totals, plus Prompts (the prompt queue) and Chats buttons — and the chat panel opens from any page. Details in Own Agent.

The Typical Path

  1. Add a server — name, host, SSH port, user, password or private key. Details in Servers.
  2. Check the connection — MultiAgent tests SSH and shows the connection status.
  3. Install Docker — one button, with the output streaming into the interface.
  4. Add an agent — the Add Agent wizard: pick the type — a standalone agent or a product (Research, Analytics, Data, Proxy, MultiAgent, chatbots, Telegram, UI, CRM); the name, server, image and port are filled in for you. Details in Agents.
  5. Deploy the agent or product — MultiAgent generates compose.yml and starts the container on the server. Products deploy together with a linked agent: by default the server’s first standalone agent is used, and if there is none, a new one is created by the same wizard.
  6. Enable the route — install Nginx from the server’s tab, and the agent immediately gets its own subfolder {serverHost}/{slug}/ on the server. Details in Nginx Routing.
  7. Set up the home page — MultiAgent builds a static card board at the server’s root: agents, apps, tools, a Git card and a “Sign out” button. No separate container and no separate login — Nginx is enough. Details in Server Home Page.
  8. Turn on central login — the “Authorization” (Authelia) tab on the server page: a visitor signs in once and gets access to every protected app on the server. It needs HTTPS — a local certificate from the Nginx tab, or an upstream reverse proxy with the address template https://{serverHost}/{slug}. Details in Servers.
  9. Open the chats — chats from all agents appear in one shared list. Details in Chats.

Interface Language

The entire interface is in English — screens, labels and messages. The language is a property of the product edition, so there is nothing to configure per user.

← Back to documentation