dLazy Ecommerce Visual Skills

dLazy's Ecommerce Visual Skills teach your agent to make the images and videos a product listing needs: from a garment flat-lay to an on-model shot, a full variant set, a detail page, a short video and a main image checked against marketplace rules. Each skill is a set of steps the agent follows, with bundled scripts for the parts that should be code, such as batching, retries, cost caps, compliance checks and a quality-check loop. What its skills teach your agent - Put products on people: a garment from its flat-lay, shoes, bags and accessories worn by a model, and up to 8 separate items styled into one look - Turn one image into many: swap the model or background, build a multi-angle variant set, and lay out detail-page modules with typeset copy - Make single edits: ghost-mannequin 3D shots, clean flat-lays pulled from any photo, fabric swaps, macro detail shots, new backgrounds, retouching, texture rebuilds and removing promotional text - Make video: a short clip from a main image, a short-form ad cut from a storyboard with subtitles, and UGC-style talking-head videos - Work at scale: run a CSV of products through one visual style, share a brand kit across every skill, and check AI images for defects before launch - Get listings right: check main images against Amazon, TikTok Shop, Temu, Shopee, Shopify and Taobao rules and fix them, plan A/B main-image tests, and localize an asset set for other regions How to install it In Claude Code, run /plugin marketplace add dlazy-ai/ecommerce-skills, then /plugin install ecommerce-skills@ecommerce-skills. The skills alone also install with npx skills add https://github.com/dlazy-ai/ecommerce-skills --all. What it needs Generating images needs a backend key. dLazy is the default (dlazy login, or DLAZY_API_KEY), and OpenAI, Gemini, fal, Replicate and Volcengine Ark work with your own key. The scripts need Node 20 or later, the compliance check needs Python with Pillow, and video assembly needs ffmpeg. Every script has a dry run that shows what would be sent and its estimated cost without generating anything. Good to know - The skill instructions and their trigger phrases are written in Chinese; the README is in Chinese and English. - Generation is paid to the backend you choose; dLazy charges credits per image. - With dLazy, prompts go to api.dlazy.com and your input images are uploaded to files.dlazy.com. With another backend, they go to that provider instead. - The UGC video skill says not to present generated videos as real buyer reviews and to label AI content where a platform requires it. The watermark skill says not to remove other people's copyright watermarks, brand marks or required labels.

• “Put the sweater from this flat-lay on a model, matching this pose reference.”

  Copy

• “Turn this product photo into a set of hero, lifestyle and close-up shots.”

  Copy

• “Check this main image against Amazon's rules and fix what would get it rejected.”

  Copy

• “Run every product in skus.csv through one consistent studio style.”

  Copy

• “Check this AI-generated image for defects before we launch it.”

  Copy

• “Make a short product video ad with subtitles from these selling points.”