What problem does it solve? It turns natural-language requirements, PRDs, screenshots, or uploaded assets into runnable, previewable web applications, and routes follow-up questions about an existing app's content, source code, errors, or operational metrics to the sandbox that actually holds the artifact. ## Core Features & Use Cases - Web app generation: Build websites, H5 pages, admin dashboards, data boards, and interactive prototypes from natural language, documents, or uploaded assets, choosing between jspage, html, and fullstack architecture types. - Web app editing: Apply page-level or global changes—feature additions, style adjustments, bug fixes—to an existing application identified by its app ID. - Artifact Q&A: Answer questions about a generated app's page content, source code, runtime errors, and operational data (traffic, users, publish status) by routing the query to app_builder_agent, which alone can read the sandbox. - Use Case: A user asks for a sales dashboard prototype with mock data; the skill selects the jspage architecture, assembles the full prompt with any uploaded image URLs, and calls app_builder_agent to produce a previewable app. ## Quick Start Ask the assistant to create a web application from your requirement, for example: build an interactive data dashboard prototype with mock sales data.
