Step 6 - Packaging the Agents

Packaging is a critical step in the lifecycle of an AI agent and its associated workflows in the HCL UnO Agentic AI Builder. When you package an agent, all its components: agents, subagents, workflows, configurations are consolidated into a single, deployable unit. This ensures consistency, portability, and reliable execution across different environments.

Accessing the Packaging Interface

The New Package button is persistently available across major sections of the application (for example, Catalog, AI Agents, Packages, and Credentials), allowing you to initiate package creation without interrupting your workflow. Clicking this button opens the Create Package page.

The Create New Package page consists of three main sections that facilitate easy configuration and visualization of your package:

  1. Profile Settings (Left Panel): This section allows you to configure the metadata and control settings for the package.

  2. Agent Selection (Right Panel): This section allows you to select the agents and related components to include in the package.

  3. Package Visualization (Central Canvas): The central canvas reflects your selections, showing all included agents and workflows in a consolidated view.

Creating Packages

  1. Click the New Package button, the Create new package page appears.
  2. In the left panel define the Profile settings.
    Field Description
    *Indicates mandatory fields
    Tags*

    Tags serve as keywords or labels that you can assign to your packages. They categorize and organize packages, making them easier to find and manage. You can add up to 5 tags to categorize or filter the package in the catalog.

    You can either select tags from existing list or enter new the tag and click enter.
    Name* Enter a unique name for the package’s purpose or content.
    Description*

    A brief explanation of what the package contains, its overall functionality, and any significant changes or features included in this specific version.

    LLM (optional if 1 agent)

    Overrides agents LLM    		 Select this checkbox to specify a different LLM (Large Language Model) at the package level instead of using the one defined within individual agents.
    Note: This setting is optional if the package contains only one agent, but becomes mandatory when there are two or more agents to ensure consistent LLM behavior across all agents.
    LLM type

    Appears only if override is checked. Select the LLM provider and model (for example, OpenAI → GPT_4O_MINI).

    For details on fields, refer to the Configuring Parameters field description table.
  3. In the right panel choose the agents, workflows, and related components you want to include in the package. This lists all available agents grouped by library and categories such as Sales > Domains and so on.
  4. As you select components from the right panel, they will appear in the New Package box in the central canvas.
  5. Verify the visual representation to confirm which agents and workflows are being included in your package.
  6. After entering all the details, click the Save button.

Result: The new package created will be displayed on the Packages page.

Remember:

  • Packages must contain at least one agent.
  • If only one agent is included and no override is selected, the agent’s default LLM will be used.
  • Only agents with Ready status appear in this list for selection.
  • Saved packages can be versioned and reused in other deployments.
  • Once the package is ready, change the status to Ready from In-progress.