SnipToss – Screenshot & Annotate for AI

Capture part of a web page, mark what to change and write notes, then paste one image into ChatGPT, Claude, Codex or a chat app.

Point at it. Say what to change. Paste one image into AI. Telling an AI what's wrong with a web page is hard. "The second button on the right, a bit lower, make it blue" takes a paragraph, and the AI still guesses. With SnipToss you show it instead: capture the page, mark exactly what you mean, write a short note for each mark, and paste one image into ChatGPT, Claude, Codex or any chat app. HOW IT WORKS 1. Start – press Alt+Shift+S (⌥⇧S on Mac), click the SnipToss icon in the toolbar and choose "Capture & annotate", or right-click the page and choose "Capture and annotate with SnipToss". 2. Select – click an element to take the whole thing: a card, a button, a paragraph. Scroll to pick a bigger or smaller one, or drag to select any area. 3. Mark – draw a rectangle, ellipse, arrow or pen stroke and write what to change. Every mark gets a number. 4. Finish – press Enter. One image goes to the clipboard: your screenshot with the numbered marks, and your notes listed below it. The cursor goes back to the box you were typing in, so you can paste right away. ANNOTATION TOOLS • Rectangle, ellipse, arrow and pen, each with its own note • Numbered pin: click anywhere to place a note • Text on the image, and mosaic to hide sensitive information • Overall comment for notes that aren't about one spot • Six mark colors, line and text sizes, undo • A magnifier with the pixel position and color under the cursor • A one-key shortcut for every tool NOTES THE AI CAN FOLLOW • Every note is numbered to match its mark on the image • Tag each note Change, Remove, Add or Question • Each note says where its mark is, in pixel coordinates of the screenshot • Notes are drawn under the picture, so nothing is lost in apps that only take the image when you paste, such as Claude Code and Codex • Prefer text? Turn off "Include notes in the image" and the notes go to the clipboard as separate text • Add the same line to every capture, for example "Keep the existing style and only change what's marked." SAVE OR SHARE • Enter, or a double-click on the selection, copies the image to the clipboard • Ctrl+S (⌘S on Mac) saves it as a PNG file • If a page blocks the clipboard, SnipToss copies the image from its own popup instead PRIVACY • Everything happens in your browser. Screenshots and notes are never uploaded, and SnipToss makes no network requests • No account, no sign-up, no analytics, no ads • SnipToss only sees a page when you start a capture on it • Minimal permissions: activeTab, scripting, storage and contextMenus. No access to your browsing history or to other tabs GOOD TO KNOW • SnipToss captures the visible part of the page; it doesn't stitch long scrolling pages • Chrome doesn't let extensions capture its own pages, such as chrome:// pages and the Chrome Web Store • To capture files opened from your computer, turn on "Allow access to file URLs" for SnipToss on its details page • You can change the shortcut at chrome://extensions/shortcuts • Interface languages: English, Spanish, German, French, Portuguese (Brazil), Japanese, Korean, Simplified Chinese and Traditional Chinese SUPPORT SnipToss is free. Made by MoyGoldenAge. Questions or ideas: jujuw3939@outlook.com If it saves you time, you can support development with a coffee: https://ko-fi.com/sheethappensco