imraywang/ray-sticker-shop

What it does

ray-sticker-shop is an Agent skill (for Claude Code or Codex) that takes a single reference photo or character image and produces a full, visually consistent animated sticker set, then packages it into the assets required by the WeChat Sticker Open Platform. It addresses the consistency and sizing problems of making GIFs one at a time: one reference image locks the character across the whole set, GIFs are output at 240×240 and automatically compressed under 500KB, and cover, icon, banner and appreciation images are generated together. The skill also checks the output against WeChat's specifications and produces listing copy plus a delivery package for clients.

Quick start

1. Prepare Python 3.9+ and Pillow, and set an OPENAI_API_KEY (also set OPENAI_BASE_URL if using a relay service). Install Pillow with pip install pillow. Without a key, the Agent can provide prompts for you to generate images in ChatGPT and put them back for assembly. (setup requirements)
2. Clone the skill into your Agent's skills directory, for example:

git clone https://github.com/imraywang/ray-sticker-shop ~/.claude/skills/ray-sticker-shop

3. In Claude Code or Codex, ask it to turn a photo into a sticker set, e.g. 「把这张猫的照片做成一套表情包。」 The skill then generates the images, assembles the GIFs and supporting assets, and applies the WeChat checks. (installation)

Cost

For 8 stickers plus 5 supporting assets, the README estimates roughly $0.7 at OpenAI's official pricing; regenerating images costs extra. Actual spending is recorded in costs.jsonl in the project directory. (cost notes)

Constraints

Only use images you have the rights to; WeChat does not accept unauthorized derivative or fan works. Approval is decided by WeChat, and the tool does not guarantee approval or revenue. The project is MIT licensed.

HotRepo's take

The repository appeared on HotRepo's new-project list for Python on 2026-11-06, gaining 110 stars and 11 forks in roughly 14 hours of that window — a fast initial pickup for a niche Agent skill. That single snapshot only supports the fact of that day's growth; it does not establish a trend. What stands out technically is the scope: it handles not just image generation but the WeChat-specific packaging and compliance checks, which is the part that usually takes the most manual work.