GitHub radar
Yingzao: Photo-to-Poster Skill for AI Agents
An agent skill that transforms photos of buildings, streets, cafés, and food into art-directed editorial posters with custom typography — not a filter, a full design pass.
Yingzao is an agent skill for Claude Code and Codex that transforms photos of buildings, streets, cafés, artifacts, and local food into art-directed editorial posters. Unlike a photo filter, the skill performs a full design pass: the agent reads the photo's composition, selects a matching visual reference from a built-in recipe library, designs a Chinese display-type system and spatial layout, then sends the original photo, reference, and typographic guide to an image-generation model in a single call. Subject extraction, color field, and typography all exist in the same visual world — not layered on top. After delivery you can optionally extend the poster into a 9-frame video storyboard with a ready-to-use prompt for a video model. Requires Python 3.10+ with Pillow, NumPy, OpenCV, and fontTools, plus an agent that supports Skills and image generation.
Why a vibe-coder should care
If you regularly create visual content around places, food, or architecture, this skill lets you hand a photo to your agent and get a designed poster back — no Photoshop, no manual prompting for image generation. The skill handles composition decisions, type design, and the generation call, then returns the result with a one-pass diagnosis and optional video extension.
How to install
Copy this and send it to your agent — Claude Code, Codex, any of them:
Install the photo-to-editorial-poster skill from https://github.com/op7418/guizang-yingzao-skill — run `npx skills add https://github.com/op7418/guizang-yingzao-skill --skill yingzao`, check dependencies with `python3 yingzao/scripts/check_dependencies.py`, then create a poster from my photo. Ask if you need any API keys.
Runs on a regular laptop — Python 3.10+ and an agent with image generation support required.
Open on GitHub▌ More finds