The terminal interface provides these primary commands, designed to work with agent execution, multi-provider selection, and session persistence.
Starts an intelligent session. The agent reads repository context, selects providers and models, plans steps, executes with permission guards, and validates results.
Produces artifacts, applications, or structured outputs from descriptions using agent planning and real tool execution.
Starts a local preview server for the current project or generated artifact.
Manages persistent work sessions. Sessions retain repository awareness, selection state, and conversation history.
Configures providers and model selection. Intelligent routing matches task requirements to capabilities.
Controls agent execution mode, tool registry access, and permission policies. Supports autonomous loops with bounded retry.
Exports and imports session state and project metadata safely.
superx run --agent analyst
superx session --continue lastNext: Architecture