Consistent Social Media Cover Agent Skill

Toss in an article, three rounds of Q&A, directly output image-generating cover prompts. No need to understand design, no need to write prompts yourself, frame fixed at 3:4, and face consistency can be maintained👇 👹The entire workflow is as follows: 1. Skill installation and initial configuration (save a front-face photo of yourself + confirm the image generation model) 2. Toss the article content to the agent, it automatically reads and extracts 3. First round: recommend composition styles based on content (10 types to choose from) + automatically think of candidate titles 4. Second round: face reference image + UI screenshot/product images, etc. 5. Third round: expression, background color tone, font, text color all selected at once; if lazy, leave it all to the model 6. Output complete prompts: 3:4 composition, safe area, multi-reference writing for image 1 face + image 2 material, all included 7. Take it to Jimeng / Nano Banana / GPT-Image to generate images (if your agent is configured with an image generation API, you can have it generate directly in the conversation) 8. Check if the title has typos; if unsatisfied with any item, change that item and regenerate 🧐Tips: Must choose a model that supports multiple reference images, otherwise face consistency cannot be maintained The skill has built-in 8 sets of example prompts reverse-engineered from real hit covers; the model will automatically use them as references, directly maximizing image specificity.

Machine-translated — the original is what produced this result.

✨ Generate Save Copy prompt Share Original post Open in Hitboard