planning-visual-creative-orchestrator

Multi-domain visual design generation is coordinated by dispatching isolated Claude Code Task subagents for each pass, applying per-pass palettes, directions, libraries, and anti-repeat constraints to ensure unique cross-domain outputs.

Playwright-based validation runs automatically after each visual concept generation pass, enforcing per-pass uniqueness and anti-repeat constraints to verify the self-contained index.html, style.css, and app.js outputs.

A central style-config.json file is required to enumerate domains, styles, and per-pass definitions, which the orchestrator reads to dispatch subagents and organize outputs under the design concepts directory.

Generating self-contained visual outputs per pass is best handled by dispatching isolated subagents that produce index.html, style.css, app.js, README, and validation metadata for each domain, style, and pass combination.

The orchestrator reads a library catalog to select domain-appropriate libraries and versions, ensuring each isolated subagent receives the correct dependencies for its specific data-vis, animation, or graphic-design pass.

Subagent-based visual orchestration requires strict anti-repeat constraints and per-pass uniqueness enforcement, as isolated agents generate independent outputs that must be validated sequentially with Playwright to prevent conceptual duplication.