Skip to main content

Workspace Setup

A workspace is where your team collaborates with AI. This guide walks you through creating a workspace, inviting members, and configuring the essentials.

Prerequisites​

Create a workspace​

  1. Open the workspace switcher at the top of the sidebar and click New workspace (this opens the Workspaces page).
  2. Fill in the Create workspace form with a name and confirm.

You are automatically assigned the workspace admin role in the new workspace.

Invite members​

There are two kinds of invite links, depending on what you want the newcomer to join:

  • Workspace invite — the newcomer joins a specific workspace at the chosen role. Use this when someone should land directly in one workspace.
  • Organization invite — the newcomer joins your organization (as Admin or Member) but is not yet placed in a workspace. After accepting, they run a short onboarding step to create their first personal workspace. Create organization invites from the org admin Members area.

To invite to a workspace:

  1. Navigate to your workspace, open Settings in the sidebar, then select the Members tab.
  2. Click Invite member.
  3. Enter the user's email address and choose a role:
    • Admin — can manage workspace settings, tools, skills, and members.
    • Member — can chat, use tools, and install personal skills.
  4. The invited user receives an email with a link to accept.

Join an existing workspace​

Click the invite link you received by email. If you do not have a CubePlex account yet, you will be prompted to create one first.

Configure models​

Before your team can chat, at least one AI model must be enabled at the organization level.

  1. Open the Admin area from the avatar menu (Admin panel), then go to Models > Model Providers.
  2. Add a provider (Anthropic, OpenAI, or a custom endpoint) and enter your API key.
  3. Enable the specific models you want available.

Models enabled at the org level are available across all workspaces.

📸 Screenshot placeholder

Capture: The Admin Models > Model Providers page with a provider being added — show the provider type selector, the API-key field, and the per-model enable toggles. Asset: /img/admin/model-providers.png

Install MCP tools​

MCP connectors let the agent interact with external services.

  1. Open the MCP page from the workspace sidebar.
  2. Browse the connector catalog and click Install on the one you want.
  3. Provide authentication credentials (a static credential, OAuth, or none, depending on the connector).
  4. Choose which workspace members should have access via Grants.

Once enabled and credentialed, the tools are available to the agent in conversations. See the MCP Tools guide for a full walkthrough.

Install skills​

Skills extend what the agent can do. To add skills to your workspace:

  1. Open the Skills page from the workspace sidebar, or use the / command inside a conversation to discover skills.
  2. Browse built-in skills, org-uploaded skills, or remote registries.
  3. Click Install to make a skill available in the workspace.

See the Skills guide for details.

Set up automation (optional)​

If you want the agent to run tasks on a schedule or in response to events:

  1. Open the Scheduled Tasks page (for cron, interval, or one-shot runs) or the Triggers page (for webhook URLs that start an agent run) from the workspace sidebar.
  2. Create a scheduled task or an event trigger.
  3. Configure the prompt, model, and any tools the automated run should use.

See the Automation guides for details.

Next steps​

Your workspace is ready. Here is where to go from here:

  • Conversations — Start chatting with full context on features.
  • Memory — Set up shared memory so the agent remembers team knowledge.
  • Members & roles — Manage organization-wide membership.
  • Cost tracking — Monitor API usage across your organization.