All Smart Components are created through the UnqorkAI Build Agent. There is no manual assembly. Creators describe what they need in chat and the agent produces the component.
Workspace Scope
The Build Agent always operates in the context of the active workspace. When creating or editing a Smart Component, the agent works from the workspace selected in the top-left dropdown, not from a specific application inside it. If the active workspace changes, the agent context resets.
Entry Points
The Build Agent chat is available from two locations for creating a Smart Component.
| Entry point | When to use |
|---|---|
| Module Builder | Creating a component for an immediate need while building a module. The component is added to the canvas after creation. |
| Smart Component Library | Creating a component before placing it in a module, or when building components for shared use across the environment. |
In both cases, the experience is the same: describe the component in chat and the agent builds it.
Scoping a Build
The Build Agent handles one task at a time. A task covers exactly the scope of a single instruction or request. For complex builds, such as a multi-page application, break the work into phases and give the agent one phase per task rather than the full requirements at once. This keeps each task focused and reduces the risk of hitting context limits.
The agent moves through two internal modes during a build: a plan mode (lower token usage) and a build mode (higher token usage, because it needs full knowledge of Centauri). The agent switches between them automatically.
When the agent presents a plan and asks for approval, a Approve All option is available to let all subsequent steps proceed without manual confirmation at each one.
Reuse Before Build
Before building a new component, the agent checks whether an existing component already covers the need. This happens automatically. Creators do not need to search the library manually.
The agent checks for reuse in the following order:
- Adopt an existing component: If a component in the environment already serves the need, the agent presents it as a match for adoption.
- Build new: If no existing component fits, the agent creates one.
When an existing component is presented as a match, review it before adopting. If it covers most of the need but not all, describe the gap to the agent. The agent can extend the existing component or build a separate one depending on the scope of the difference.
What the Agent Produces
When the agent builds a Smart Component, it generates the following:
- The component's generated code.
- A typed input and output interface declaring what the component accepts and returns.
- A name and description based on the chat description.
The component is added to the Smart Component Library immediately and becomes available in the Component Library section of the component tray.
Editing After Creation
After a component is created, open it from the Smart Component Library or select Edit Base on the canvas. Both paths open the same component definition in the component editor. From there, describe the needed change in the Build Agent chat and the agent applies it.
Because a Smart Component is a shared, environment-level definition, any change takes effect in every module that uses it. This applies whether the edit was made from the Smart Component Library or from a specific module using Edit Base. Describe the intended change clearly, noting whether it should affect existing behavior or add something new.
Changelog
| Date | Change |
|---|---|
| 2026-07-29 | Added Workspace Scope and Scoping a Build sections (EN-7975). |
| 2026-07-28 | Initial draft (EN-7975). |