偷师是一款图片提示词反推工具，适合 AI 绘画、电商设计、海报参考、视觉创作学习使用。
你可以在网页图片上点击悬浮按钮，或上传本地图片，让插件推测图片可能对应的 AI 生成提示词。插件支持中文与英文提示词、图片比例识别、风格标签、历史资产收藏，并可一键带提示词跳转到 PixPix 生成同款。
主要功能：
- 网页图片悬浮反推
- 本地图片上传反推
- 中文 / 英文 / JSON 提示词切换
- 自动识别图片类型、画面比例和风格标签
- 复杂海报、信息图、拼贴图自动精修补全
- 本地资产收藏与分类
- 一键跳转 PixPix 生成同款
- 用户自带智谱 API Key，Key 只保存在本机
V2.1.0更新说明：
1、全新网页内浮窗体验：反推结果直接显示在当前页面，可拖拽移动，不再固定占用浏览器右侧。
2、新增智能截图反推：按 Alt + Q 自动吸附图片区域，可像微信截图一样手动框选后反推。
3、新增本地资产库：反推过的图片、提示词和分类会自动保存，方便后续查看和复用。
4、提升复杂图片识别效果：海报、信息图、拼贴图、商品图等会更完整识别文字、版式、元素和风格细节。
5、优化多网站取图能力：遇到无法右键、无法复制或无法直接抓图的网站，可用截图反推兜底。
隐私说明：
插件不会收集或上传你的智谱 API Key，不会上传你的历史记录，不会读取无关网页内容。只有当你主动点击反推或上传图片时，插件才会把该图片发送到智谱 API 用于生成提示词。
提示：
结果由 AI 推测，仅供学习参考，不代表原作真实提示词。
MasterStudy AI is an image prompt reverse-engineering tool for AI art, e-commerce design, poster references, and visual creation learning.
You can click the floating button on any webpage image, or upload a local image, and the extension will infer the possible AI generation prompt behind it. The extension supports Chinese and English prompts, aspect ratio detection, style tags, history assets, local collections, and one-click prompt transfer to PixPix to generate a similar image.
Key Features:
· Reverse prompts from webpage images
· Reverse prompts from uploaded local images
· Switch between Chinese / English / JSON prompts
· Automatically detect image type, aspect ratio, and style tags
· Auto-refine complex posters, infographics, and collage images
· Local asset collection and category management
· One-click jump to PixPix to generate a similar image
· Bring your own Zhipu API Key. Your Key is stored only on your device.
V2.1.0 Update Notes:
1. New in-page floating window experience: reverse prompt results now appear directly on the current webpage and can be dragged freely, instead of being fixed on the right side of the browser.
2. New smart screenshot reverse prompt feature: press Alt + Q to automatically snap to the image area, manually adjust the selection like a screenshot tool, and then reverse the prompt.
3. New local asset library: reversed images, prompts, and categories are automatically saved for later viewing and reuse.
4. Improved complex image recognition: posters, infographics, collage images, and product images can now capture text, layout, elements, and style details more completely.
5. Better compatibility across websites: when images cannot be right-clicked, copied, or directly captured, smart screenshot reverse prompt can be used as a fallback.
Privacy:
The extension does not collect or upload your Zhipu API Key, does not upload your history, and does not read unrelated webpage content. Only when you actively click reverse prompt or upload an image will the selected image be sent to the Zhipu API to generate a prompt.
Note:
Results are inferred by AI and are for learning/reference only. They do not represent the original creator’s actual prompt.
Data sourced from the Chrome Web Store; Extenshi is not affiliated with or endorsed by it. Details