article-to-cover

Overview

Art-director-level poster design skill. Analyzes text input, anchors style direction, plans visual hierarchy, and outputs structured AI generation instructions. Two modes: creative direction from scratch (no reference image), or reference-image-based reconstruction/mimicry.

Dependencies

• engine: meitu-ai
• user data: ~/.openclaw/visual/

Core Workflow

Step 1: Load Context

1. Analyze user input — Read the provided text (article, chat transcript, or design brief). Synthesize core message into headline + subtitle candidates.
2. User preferences — If ~/.openclaw/visual/memory/global.md exists, read for style preferences. If contexts/poster.md exists, read for poster-specific preferences.
3. Experience log — If ~/.openclaw/visual/journal/knowledge.yaml exists, scan for entries related to this industry or similar tasks.
4. Brand assets — If user specifies a brand, read ~/.openclaw/visual/assets/brands/{brand}/: